diff --git a/02_activities/assignments/a1_sampling_and_reproducibility.ipynb b/02_activities/assignments/a1_sampling_and_reproducibility.ipynb index 873f5985..e3affafa 100644 --- a/02_activities/assignments/a1_sampling_and_reproducibility.ipynb +++ b/02_activities/assignments/a1_sampling_and_reproducibility.ipynb @@ -16,7 +16,21 @@ "cell_type": "markdown", "id": "4ea73db3", "metadata": {}, - "source": [] + "source": [ + "There are three stages of sampling in this simulation:\n", + "\n", + "Stage 1 — Infection sampling (primary)\n", + "Using np.random.choice(ppl.index, size=int(len(ppl) * ATTACK_RATE), replace=False), a random sample of 100 people (10% of 1000) is selected to be infected. The sampling frame is all 1000 attendees (200 at weddings, 800 at brunches). This is uniform random sampling without replacement, which follows a hypergeometric distribution — each person has equal probability of being chosen.\n", + "\n", + "Stage 2 — Primary contact tracing\n", + "Using np.random.rand(sum(ppl['infected'])) < TRACE_SUCCESS, each infected person is independently traced with probability 0.20 (TRACE_SUCCESS). The sampling frame is the 100 infected people. Each person's tracing outcome follows a Bernoulli distribution with p=0.20, so the total number traced follows a Binomial distribution (n≈100, p=0.20), yielding approximately 20 traced cases.\n", + "\n", + "Stage 3 — Secondary contact tracing (threshold-based escalation)\n", + "This is not random sampling, but a deterministic rule: if any event has 2 or more traced people (SECONDARY_TRACE_THRESHOLD = 2), then ALL infected people at that event get traced. This step introduces bias — larger events (weddings) are more likely to hit the threshold, causing disproportionate tracing of wedding-linked infections.\n", + "\n", + "Stage 4 — Monte Carlo simulation repetitions\n", + "simulate_event is called 1000 times via [simulate_event(m) for m in range(1000)]. Each repetition independently draws new random samples. This produces a distribution of outcomes rather than a single result." + ] }, { "cell_type": "markdown", @@ -30,7 +44,13 @@ "cell_type": "markdown", "id": "4cf5d993", "metadata": {}, - "source": [] + "source": [ + "With 10 repetitions, the results are highly non-reproducible. Each run produces a very different histogram because 10 samples are far too few to represent the underlying distribution — the law of large numbers has not kicked in. The plots change dramatically between runs.\n", + "\n", + "With 100 repetitions, the results are slightly more stable, but still noticeably different between runs. The general shape of the distribution begins to emerge (infection proportions cluster around 0.20, traces skew higher toward weddings), but the exact histogram varies each run.\n", + "\n", + "With 1000 repetitions, the results are visually more consistent between runs because the larger sample better approximates the true distribution. However, without a fixed random seed, the results are still not exactly reproducible — each run will differ slightly. The key takeaway is that more repetitions reduce variance but do not guarantee reproducibility on their own." + ] }, { "cell_type": "markdown", @@ -44,7 +64,11 @@ "cell_type": "markdown", "id": "77613cc3", "metadata": {}, - "source": [] + "source": [ + "I added np.random.seed(44) before running the simulation. NumPy's random functions (np.random.choice, np.random.rand) use a pseudo-random number generator (PRNG). Without a seed, the PRNG initializes from an unpredictable system state (e.g., current time), so each run produces a different sequence of random numbers and thus different results.\n", + "\n", + "By setting a fixed seed with np.random.seed(44), the PRNG always starts from the same internal state. This means every call to np.random.choice and np.random.rand produces the identical sequence of numbers on every run, resulting in the same infected individuals, same traced individuals, and the same output histogram each time the script is executed. The value 44 is arbitrary — any integer constant will work as long as it stays the same." + ] }, { "cell_type": "markdown", @@ -56,10 +80,21 @@ }, { "cell_type": "code", - "execution_count": null, + "execution_count": 9, "id": "ab8587a0", "metadata": {}, - "outputs": [], + "outputs": [ + { + "data": { + "image/png": "iVBORw0KGgoAAAANSUhEUgAAA90AAAJOCAYAAACqS2TfAAAAOnRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjEwLjYsIGh0dHBzOi8vbWF0cGxvdGxpYi5vcmcvq6yFwwAAAAlwSFlzAAAPYQAAD2EBqD+naQAAbc5JREFUeJzt3Xt8z/X///H7e+e9Zxs2zPmQ04TkkFC2JeSUqFTOhz6RkoXIR2WpkHLoo6g+CeXjkEI6EMmxUc4VC0lGNlqYMRvbnr8/fPf+9baNmffLe7hdL5f35fN5Pd/P9+v1eL2f79db971er+fbZowxAgAAAAAALufh7gIAAAAAALhREboBAAAAALAIoRsAAAAAAIsQugEAAAAAsAihGwAAAAAAixC6AQAAAACwCKEbAAAAAACLELoBAAAAALAIoRsAAAAAAIsQugEUyKxZs2Sz2bRlyxZ3l3LVdu/erZiYGP3xxx8uX/eqVavUsGFDBQQEyGazacmSJZfsf/ToUT3//POqU6eOihQpIj8/P1WrVk2DBw/Wvn37XF7fP40dO/ay9V3L7fTu3Vs2m+2yj969e1tesyTZbDbFxMRck225wx9//OH0vnp4eCgkJERt27bVxo0b3V1enmJiYmSz2dyy7cjISEVGRl62X6VKlfL8/J4+fVrS//9OdeX3UKVKldS+ffsCv37q1KmqWrWqfHx8ZLPZdPLkSZfVJl36u7d3796qVKmSS7eXX3///bdGjhypWrVqKSAgQMHBwapZs6Z69Oihn376yS01Abi+ebm7AABwt927d+vll19WZGSkS/8jzxijLl26qHr16lq6dKkCAgJUo0aNPPv/+OOPat++vYwxevrpp9WkSRP5+Phoz549mjNnju644w6dOHHCZfVdbOzYsXrooYf0wAMPWLaNK9nOiy++qAEDBjiWt23bpqeeekpjx45VVFSUo71EiRJWlepk48aNKleu3DXZljsNGjRIXbt2VWZmpnbt2qWXX35ZUVFR2rhxo26//XZ3l5fD448/rvvuu8/dZVxWs2bN9Oabb+Zot9vtbqjm8nbs2KFnnnlGjz/+uHr16iUvLy8FBga6dBuX+u598cUXNXjwYJduLz9Onz6tO++8U6dPn9Zzzz2n2267TWfPntXevXu1aNEi7dixQ3Xr1r3mdQG4vhG6AcAiR44c0fHjx9WpUye1aNHikn1PnTqljh07ys/PT7GxsU7hLjIyUv3799enn35qdcmFyi233KJbbrnFsZyWliZJqlatmu688848X3f27Fn5+fm5/OznpbZ5I6lQoYJjX5s1a6aqVauqRYsWmjZtmv773/9e1bpTU1NdHjLLlSt3XfwxpGjRotfVZ2jXrl2SpH/961+64447rvn2/3nsX0sLFy7Ub7/9pu+++87pj3uSNGTIEGVlZV3zms6ePSt/f/9rvl0ArsPl5QBcpnfv3ipSpIh+/fVXtW7dWgEBASpdurTGjx8vSdq0aZPuuusuBQQEqHr16po9e7bT67Mvr1y5cqX69Omj4sWLKyAgQB06dNDvv//u1HflypXq2LGjypUrJz8/P1WtWlX9+/dXUlJSjrp+/fVXPfbYYypVqpR8fX1VoUIF9ezZU+np6Zo1a5YefvhhSVJUVJTjks9Zs2Zdcl83bNigFi1aKDAwUHa7XU2bNtVXX33leD4mJsYRBEaMGCGbzXbJs+j//e9/lZiYqAkTJuQZIB566CGn5aVLl6pJkyay2+0KDAxUy5Ytc1wGnH3p7a5du/TYY48pODhYpUqVUt++fZWcnOzoZ7PZdObMGc2ePdvxHmRfNvvXX39p4MCBqlWrlooUKaKSJUvqnnvu0fr163PUmJ6erjFjxig8PFx+fn4KCQlRVFSUYmNjL7udgsj+zKxYsUJ9+/ZViRIlZLfblZ6ert9++019+vRRtWrVZLfbVbZsWXXo0EE///xzjvWcPHlSQ4cOVZUqVeTr66uSJUuqbdu2+vXXX53eo39eXp697dWrV+vJJ59UaGioQkJC1LlzZx05ciTH+zJ06FCFhYXJbrerefPm2rp1qypVqpSvy+OPHz+ugQMHqmzZsvLx8VGVKlU0atQopaenO/Wz2Wx6+umn9fHHHys8PFx2u1233Xabvvzyyyt7Y/8hOygePHjQ0fbtt9+qRYsWCgoKkt1uV7NmzbRq1Sqn12V/9rZt26aHHnpIxYoVcwSprKwsTZ06VfXq1ZO/v78jkC5dutRpHQsWLFCTJk0UEBCgIkWKqHXr1tq+fXuu28n2wAMPqGLFirmGo8aNG6t+/fqOZWOMpk2b5qijWLFieuihh3J83xhjNGHCBFWsWFF+fn6qX7++li1bdiVvY4Hk9fnI72XtF8u+heDNN9/UpEmTVLlyZRUpUkRNmjTRpk2bnNbfvXt3SRfes4tv48jP+EtX992b2+XlaWlpGjlypCpXriwfHx+VLVtWTz31VI7L3rMvq1++fLnq168vf39/1axZUx9++OFl36O///5bklS6dOlcn/fwcP5P58v9WyDlfQtEbrcUZNe+aNEi3X777fLz89PLL78sSfrzzz/1xBNPqHz58vLx8VGZMmX00EMP6ejRo47Xnzp1SsOGDXN6j6Kjo3XmzBmnbS9cuFCNGzdWcHCw7Ha7qlSpor59+172/QFQMIRuAC51/vx5de7cWe3atdPnn3+uNm3aaOTIkfr3v/+tXr16qW/fvlq8eLFq1Kih3r17a+vWrTnW0a9fP3l4eGju3LmaMmWKfvzxR0VGRjr9h9X+/fvVpEkTTZ8+XStWrNBLL72kH374QXfddZfOnz/v6Ldz5041atRImzZt0pgxY7Rs2TKNGzdO6enpOnfunNq1a6exY8dKkt555x1t3LhRGzduVLt27fLcx7Vr1+qee+5RcnKyZsyYoXnz5ikwMFAdOnTQggULJF245HXRokWSLlyuu3HjRi1evDjPda5YsUKenp7q0KFDvt7nuXPnqmPHjgoKCtK8efM0Y8YMnThxQpGRkdqwYUOO/g8++KCqV6+uzz77TM8//7zmzp2rZ5991vH8xo0b5e/v77h/d+PGjZo2bZqkC4FPkkaPHq2vvvpKM2fOVJUqVRQZGak1a9Y41pGRkaE2bdrolVdeUfv27bV48WLNmjVLTZs2VXx8/GW3czX69u0rb29vffzxx/r000/l7e2tI0eOKCQkROPHj9fy5cv1zjvvyMvLS40bN9aePXscr01JSdFdd92l9957T3369NEXX3yhd999V9WrV1dCQsJlt/3444/L29tbc+fO1YQJE7RmzRpHYMnWp08fTZkyRX369NHnn3+uBx98UJ06dcrXPbJpaWmKiorSRx99pCFDhuirr75S9+7dNWHCBHXu3DlH/6+++kpvv/22xowZo88++0zFixdXp06dcgTJ/Prtt98k/f/L+OfMmaNWrVopKChIs2fP1ieffKLixYurdevWuQavzp07q2rVqlq4cKHeffddSRcC1eDBg9WoUSMtWLBA8+fP1/333+8UPsaOHavHHntMtWrV0ieffKKPP/5YKSkpuvvuu7V79+486+3bt6/i4+P13XffObX/+uuv+vHHH9WnTx9HW//+/RUdHa17771XS5Ys0bRp07Rr1y41bdrUKci8/PLLGjFihFq2bKklS5boySef1L/+9S+nz9HlGGOUkZHh9HDHWVPpwnfdypUrNWXKFP3vf//TmTNn1LZtW8cf4qZNm6YXXnhBkjRz5kxt3LhRL774oqT8j7+rv3uNMXrggQf05ptvqkePHvrqq680ZMgQzZ49W/fcc0+OP0Dt3LlTQ4cO1bPPPqvPP/9cdevWVb9+/bRu3bpLvjdNmjSRJPXs2VNLlixxhPDc5OffgoLYtm2bnnvuOT3zzDNavny5HnzwQf35559q1KiRFi9erCFDhmjZsmWaMmWKgoODHbcdpaamKiIiQrNnz9YzzzyjZcuWacSIEZo1a5buv/9+GWMkXfgefuSRR1SlShXNnz9fX331lV566SVlZGQUuGYAl2EAoABmzpxpJJnNmzc72nr16mUkmc8++8zRdv78eVOiRAkjyWzbts3R/vfffxtPT08zZMiQHOvs1KmT07a+//57I8m8+uqrudaSlZVlzp8/bw4ePGgkmc8//9zx3D333GOKFi1qjh07lue+LFy40Egyq1evzte+33nnnaZkyZImJSXF0ZaRkWFq165typUrZ7Kysowxxhw4cMBIMm+88cZl11mzZk0TFhaWr+1nZmaaMmXKmDp16pjMzExHe0pKiilZsqRp2rSpo2306NFGkpkwYYLTOgYOHGj8/PwctRpjTEBAgOnVq9dlt5+RkWHOnz9vWrRo4TRWH330kZFk/vvf/17y9fndzsVWr15tJJmFCxc62rI/Mz179sxX3efOnTPVqlUzzz77rKN9zJgxRpJZuXLlJV8vyYwePTrHtgcOHOjUb8KECUaSSUhIMMYYs2vXLiPJjBgxwqnfvHnzjKTLvhfvvvuukWQ++eQTp/bXX3/dSDIrVqxwqrFUqVLm1KlTjrbExETj4eFhxo0bd8ntZH9eX3/9dXP+/HmTlpZmtm7daho1amQkma+++sqcOXPGFC9e3HTo0MHptZmZmea2224zd9xxh6Mt+7P30ksvOfVdt26dkWRGjRqVZy3x8fHGy8vLDBo0yKk9JSXFhIWFmS5duuTYTrbz58+bUqVKma5duzq9dvjw4cbHx8ckJSUZY4zZuHGjkWQmTpzo1O/QoUPG39/fDB8+3BhjzIkTJ4yfn1+e30sRERF57ke2ihUrGkk5Hv98D7I/TwcOHHB6XW6fj4iIiHxvt127do7l7DGuU6eOycjIcLT/+OOPRpKZN29ejnr++R1/JeN/td+9vXr1MhUrVnQsL1++PNfvsgULFhhJ5v3333fabz8/P3Pw4EFH29mzZ03x4sVN//7986wn25gxY4yPj49jnCpXrmwGDBhgdu7c6dQvv/8WXPwZzZbXmHt6epo9e/Y49e3bt6/x9vY2u3fvzrPucePGGQ8PD6cxM8aYTz/91EgyX3/9tTHGmDfffNNIMidPnrzsewHANTjTDcClbDab2rZt61j28vJS1apVVbp0aadJmIoXL66SJUs6XbKarVu3bk7LTZs2VcWKFbV69WpH27FjxzRgwACVL19eXl5e8vb2VsWKFSVJcXFxki781X/t2rXq0qWLyybbOnPmjH744Qc99NBDKlKkiKPd09NTPXr00OHDh6/o7FdB7NmzR0eOHFGPHj2cLnUsUqSIHnzwQW3atEmpqalOr7n//vudluvWrau0tDQdO3YsX9t89913Vb9+ffn5+Tne71WrVjnea0latmyZ/Pz83HKJ4oMPPpijLSMjQ2PHjlWtWrXk4+MjLy8v+fj4aN++fTnqrl69uu69994CbTu391b6/5djr127VpLUpUsXp34PPfSQvLwuP7XKd999p4CAgBy3F2Rf7nvx2eWoqCinCa9KlSqV57GWmxEjRsjb21t+fn5q0KCB4uPj9d5776lt27aKjY3V8ePH1atXrxxnbO+77z5t3rw5x2WsF49N9mXZTz31VJ41fPPNN8rIyFDPnj2dtuPn56eIiAinKywu5uXlpe7du2vRokWOM7eZmZn6+OOP1bFjR4WEhEiSvvzyS9lsNnXv3t1pG2FhYbrtttsc29i4caPS0tLy/F7Kr7vuukubN292egwcODDfr3eldu3aydPT07F88Wc2L/kdfyu+e7OvXLj4cvuHH35YAQEBOY6DevXqqUKFCo5lPz8/Va9ePV/HwYsvvqj4+Hh9+OGH6t+/v4oUKaJ3331XDRo00Lx58yRZ+29B3bp1Vb16dae2ZcuWKSoqSuHh4Xm+7ssvv1Tt2rVVr149p/Fp3bq1bDab4zPdqFEjSRe+kz755BP9+eefBaoTQP4xkRoAl7Lb7fLz83Nq8/HxUfHixXP09fHxcUyO9U9hYWG5tmVf5peVlaVWrVrpyJEjevHFF1WnTh0FBAQoKytLd955p86ePStJOnHihDIzM106ydKJEydkjMn1fr8yZcpI0iUvR8xLhQoVtG/fPp05c0YBAQGX7Hupew7LlCmjrKwsnThxwmnCquygkc3X11eSHO/VpUyaNElDhw7VgAED9Morryg0NFSenp568cUXncLrX3/9pTJlyuS45/FayO29GDJkiN555x2NGDFCERERKlasmDw8PPT444877fdff/3l9B/nV+py7232eJUqVcqpn5eXV47X5ubvv/9WWFhYjntCS5YsKS8vrxyft9zW6evrm6+xlqTBgwere/fu8vDwUNGiRVW5cmXHtrMvub74DwD/dPz4cafP8MVj89dff8nT0zPX4zxb9nayw8HFLvcZ69u3ryZOnKj58+erf//++uabb5SQkOB0afnRo0dljMkxLtmqVKki6f+PX17fS/kVHByshg0b5ru/lQr6fZDf8ffw8HD5d+/ff/8tLy+vHCHeZrM5/fuQ7WqPg1KlSqlPnz6Oz8y6devUpk0bDR48WI899phl/xZIuX+f/fXXX5d9P48eParffvtN3t7euT6fPedJ8+bNtWTJEv3nP/9x3GN/6623atSoUXrssccKVDOASyN0Ayh0EhMTc22rWrWqJOmXX37Rzp07NWvWLPXq1cvRJ/ve02zFixeXp6enDh8+7LLasoNbbvf6Zk+eFRoaesXrbd26tVasWKEvvvhCjz766CX7Zv/HZF41eHh4qFixYldcQ17mzJmjyMhITZ8+3ak9JSXFablEiRLasGGDsrKyrnnwzm2Sojlz5qhnz56O+0azJSUlqWjRoo7lEiVKuPQzcrHs8Tp69KjKli3raM/IyMjXf5SHhITohx9+kDHGaT+PHTumjIyMAn3eLqVcuXJ5hsPsbU2dOjXPmbgvDrEXj02JEiWUmZmpxMTEPCeryt7Op59+ekVnk7PVqlVLd9xxh2bOnKn+/ftr5syZKlOmjFq1auW0DZvNpvXr1ztC5z9lt2WPX17fS1b+lrSfn1+Oe5WlC59hV497fuR3/DMzM13+3RsSEqKMjAz99ddfTsHbGKPExMQ8/0DjKs2bN1erVq20ZMkSHTt27Ir+Lcj+Q3R6errTZy23iT+l3L/P8vM9FRoaKn9//zwnjPvnZ6Zjx47q2LGj0tPTtWnTJo0bN05du3ZVpUqVHPe1A3AdLi8HUOj873//c1qOjY3VwYMHHbP1Zv8HycX/ofzee+85Lfv7+ysiIkILFy7M8z9u/rme/JwBCQgIUOPGjbVo0SKn/llZWZozZ47KlSuX47LA/OjXr5/CwsI0fPjwPC/1y56YrUaNGipbtqzmzp3rmBhHunC542effeaY0fxK5XUWyGaz5Xivf/rppxwzpbdp00ZpaWmXnfn9Ss42XY3c6v7qq69yvL9t2rTR3r17c0y85SrNmzeXpBwTK3366af5mrioRYsWOn36tJYsWeLU/tFHHzmev1aaNWumokWLavfu3WrYsGGuDx8fn0uuo02bNpKU4484/9S6dWt5eXlp//79eW7ncvr06aMffvhBGzZs0BdffKFevXo5XVLdvn17GWP0559/5rr+OnXqSLowe7ufn1+e30tWqlSpkn766Sentr1791p+C0te8jv+Vnz3Zn/O58yZ49T+2Wef6cyZMy47Do4ePZrrBHeZmZnat2+f7Ha7ihYtekX/FmT/Yebisfziiy/yXVebNm20evXqS459+/bttX//foWEhOQ6Nrn9gcjX11cRERF6/fXXJSnHrwMAcA3OdAModLZs2aLHH39cDz/8sA4dOqRRo0apbNmyjvsfa9asqVtuuUXPP/+8jDEqXry4vvjiC61cuTLHuiZNmqS77rpLjRs31vPPP6+qVavq6NGjWrp0qd577z0FBgaqdu3akqT3339fgYGB8vPzU+XKlfO89HfcuHFq2bKloqKiNGzYMPn4+GjatGn65ZdfNG/evAL9PnRwcLA+//xztW/fXrfffruefvppNWnSxHEP8pw5c7Rz50517txZHh4emjBhgrp166b27durf//+Sk9P1xtvvKGTJ086fqLtStWpU0dr1qzRF198odKlSyswMFA1atRQ+/bt9corr2j06NGKiIjQnj17NGbMGFWuXNkpND722GOaOXOmBgwYoD179igqKkpZWVn64YcfFB4e7jiDn9d2XK19+/aaNWuWatasqbp162rr1q164403clyiGR0drQULFqhjx456/vnndccdd+js2bNau3at2rdvn+O3eq/Urbfeqscee0wTJ06Up6en7rnnHu3atUsTJ05UcHDwZa8K6Nmzp9555x316tVLf/zxh+rUqaMNGzZo7Nixatu2bYHvRS+IIkWKaOrUqerVq5eOHz+uhx56SCVLltRff/2lnTt36q+//rpkmJaku+++Wz169NCrr76qo0ePqn379vL19dX27dtlt9s1aNAgVapUSWPGjNGoUaP0+++/67777lOxYsV09OhR/fjjjwoICHD8jFJeHnvsMQ0ZMkSPPfaY0tPTc9wL3KxZMz3xxBPq06ePtmzZoubNmysgIEAJCQnasGGD6tSpoyeffFLFihXTsGHD9Oqrrzp9L8XExFzR5eUF0aNHD3Xv3l0DBw7Ugw8+qIMHD2rChAkuu0/6Sl3J+Lv6u7dly5Zq3bq1RowYoVOnTqlZs2b66aefNHr0aN1+++3q0aOHS/bx448/1nvvvaeuXbuqUaNGCg4O1uHDh/XBBx9o165deumllxx/WMrvvwVt27ZV8eLF1a9fP40ZM0ZeXl6aNWuWDh06lO+6smeAb968uf7973+rTp06OnnypJYvX64hQ4aoZs2aio6O1meffabmzZvr2WefVd26dZWVlaX4+HitWLFCQ4cOVePGjfXSSy/p8OHDatGihcqVK6eTJ0/qrbfekre3tyIiIlzyPgK4iPvmcANwPctr9vKAgIAcfSMiIsytt96ao/3imXWz17lixQrTo0cPU7RoUePv72/atm1r9u3b5/Ta3bt3m5YtW5rAwEBTrFgx8/DDD5v4+PgcM0xn93344YdNSEiI8fHxMRUqVDC9e/c2aWlpjj5TpkwxlStXNp6enkaSmTlz5iX3f/369eaee+4xAQEBxt/f39x5553miy++cOpzJbOXZ0tMTDQjRowwt956q7Hb7cbX19dUrVrV9O/f3/z8889OfZcsWWIaN25s/Pz8TEBAgGnRooX5/vvvnfpkz5r7119/ObXnNmvujh07TLNmzYzdbnealTk9Pd0MGzbMlC1b1vj5+Zn69eubJUuW5Jhd2JgLMwS/9NJLplq1asbHx8eEhISYe+65x8TGxl52O5dzqdnLL56t15gLs07369fPlCxZ0tjtdnPXXXeZ9evX5zrz84kTJ8zgwYNNhQoVjLe3tylZsqRp166d+fXXXx19Lv5s5bXt7Dr/OSNzWlqaGTJkiClZsqTx8/Mzd955p9m4caMJDg52mkk9L3///bcZMGCAKV26tPHy8jIVK1Y0I0eOdPoMZ9f41FNP5Xh9XrNg/9OVfF7Xrl1r2rVrZ4oXL268vb1N2bJlTbt27ZzGJq/PnjEXZruePHmyqV27tvHx8THBwcGmSZMmOY6hJUuWmKioKBMUFGR8fX1NxYoVzUMPPWS+/fbbHNvJTdeuXY0k06xZszz35cMPPzSNGzd2HMu33HKL6dmzp9myZYujT1ZWlhk3bpwpX7688fHxMXXr1jVffPFFgWcRz01ux2RWVpaZMGGCqVKlivHz8zMNGzY033333VXPXp7bGOf3821M/sbfmKv77s3r+2XEiBGmYsWKxtvb25QuXdo8+eST5sSJE5fc72z5ed92795thg4daho2bGhKlChhvLy8TLFixUxERIT5+OOPc/TPz78FxlyYIb5p06YmICDAlC1b1owePdp88MEHuc5entdn5dChQ6Zv374mLCzMeHt7mzJlypguXbqYo0ePOvqcPn3avPDCC6ZGjRqOY6tOnTrm2WefNYmJicYYY7788kvTpk0bU7ZsWePj42NKlixp2rZta9avX3/J9wZAwdmM+ce1iQDgRrNmzVKfPn20efPmQjPhEGCV2NhYNWvWTP/73//UtWtXd5cDAAAswuXlAABYbOXKldq4caMaNGggf39/7dy5U+PHj1e1atXUuXNnd5cHAAAsROgGAMBiQUFBWrFihaZMmaKUlBSFhoaqTZs2GjduXI6f2AMAADcWLi8HAAAAAMAi/GQYAAAAAAAWIXQDAAAAAGARQjcAAAAAABZhIjVJWVlZOnLkiAIDA2Wz2dxdDgAAAACgkDPGKCUlRWXKlJGHR97nswndko4cOaLy5cu7uwwAAAAAwHXm0KFDKleuXJ7PE7olBQYGSrrwZgUFBbm5GgAAAABAYXfq1CmVL1/ekSfzQuiWHJeUBwUFEboBAAAAAPl2uVuUmUgNAAAAAACLELoBAAAAALAIoRsAAAAAAItwTzcAAABQiGVmZur8+fPuLgO46Xh7e8vT0/Oq10PoBgAAAAohY4wSExN18uRJd5cC3LSKFi2qsLCwy06WdimEbgAAAKAQyg7cJUuWlN1uv6r/6AdwZYwxSk1N1bFjxyRJpUuXLvC6CN0AAABAIZOZmekI3CEhIe4uB7gp+fv7S5KOHTumkiVLFvhScyZSAwAAAAqZ7Hu47Xa7mysBbm7Zx+DVzKtA6AYAAAAKKS4pB9zLFccgoRsAAAAAAItwTzcAAABwHYmPj1dSUtI1215oaKgqVKhg6TYSExPVo0cPxcbGytvb27IZ22fNmqXo6Gi3zghvjFH//v316aef6sSJE9q+fbvq1avntnqupZiYGC1ZskQ7duzIs0/v3r118uRJLVmyRJIUGRmpevXqacqUKdekRisQugEAAIDrRHx8vGrUCFdaWuo126afn1179sTlO3hfHJryY/LkyUpISNCOHTsUHBxcwEqdVapUSdHR0YqOjna0PfLII2rbtq1L1l9Qy5cv16xZs7RmzRpVqVJFoaGhbq3n+eef1+eff664uDhHW1xcnGrVqqXu3bvr448/drR//PHH6tu3r06cOKEiRYpck/oWLVokb2/va7ItqxC6AQAAgOtEUlKS0tJSFR4+R3Z7uOXbS02NU1xcdyUlJVl6tnv//v1q0KCBqlWrZtk2pAuzUWfPSO0u+/fvV+nSpdW0adM8+5w7d04+Pj7XpJ6oqCi9/vrrSkxMVFhYmCRpzZo1Kl++vFavXu3Ud82aNbrjjjuuWeCWpOLFi1+zbVmFe7oBAACA64zdHq7AwPqWP1wR7CMjI/XMM89o+PDhKl68uMLCwhQTE+N4vlKlSvrss8/00UcfyWazqXfv3pKk5ORkPfHEEypZsqSCgoJ0zz33aOfOnU7rXrp0qRo2bCg/Pz+Fhoaqc+fOjm0ePHhQzz77rGw2m2MyrFmzZqlo0aJO65g+fbpuueUW+fj4qEaNGk5ndqULE2l98MEH6tSpk+x2u6pVq6alS5c6nj9x4oS6deumEiVKyN/fX9WqVdPMmTNzfS969+6tQYMGKT4+XjabTZUqVXLU+/TTT2vIkCEKDQ1Vy5YtJUlr167VHXfcIV9fX5UuXVrPP/+8MjIynN7bQYMGKTo6WsWKFVOpUqX0/vvv68yZM+rTp48CAwN1yy23aNmyZXmOz1133SVvb2+tWbPG0bZmzRo99dRTSklJ0W+//ebUHhUVle/xGT9+vEqVKqXAwED169dPaWlpTs9nZmZqyJAhKlq0qEJCQjR8+HAZY5z6REZGOl2tUKlSJY0dO1Z9+/ZVYGCgKlSooPfff9/pNbGxsapXr578/PzUsGFDLVmyRDabzXFZ+5WMmSsQugEAAABYavbs2QoICNAPP/ygCRMmaMyYMVq5cqUkafPmzbrvvvvUpUsXJSQk6K233pIxRu3atVNiYqK+/vprbd26VfXr11eLFi10/PhxSdJXX32lzp07q127dtq+fbtWrVqlhg0bSrpwSXK5cuU0ZswYJSQkKCEhIde6Fi9erMGDB2vo0KH65Zdf1L9/f/Xp0yfHGd6XX35ZXbp00U8//aS2bduqW7dujjpefPFF7d69W8uWLVNcXJymT5+e5yXjb731lsaMGaNy5copISFBmzdvdnqPvLy89P333+u9997Tn3/+qbZt26pRo0bauXOnpk+frhkzZujVV1/N8d6Ghobqxx9/1KBBg/Tkk0/q4YcfVtOmTbVt2za1bt1aPXr0UGpq7rckBAQEqFGjRk77vHbtWrVo0ULNmjVztB86dEi///67oqKi8jU+n3zyiUaPHq3XXntNW7ZsUenSpTVt2jSnbU+cOFEffvihZsyYoQ0bNuj48eNavHhxrnVe/LqGDRtq+/btGjhwoJ588kn9+uuvkqSUlBR16NBBderU0bZt2/TKK69oxIgRTq+/kjFzCQOTnJxsJJnk5GR3lwIAAACYs2fPmt27d5uzZ886tW/dutVIMg0abDWRkcbyR4MGF7a3devWfNfeq1cv07FjR8dyRESEueuuu5z6NGrUyIwYMcKx3LFjR9OrVy/H8qpVq0xQUJBJS0tzet0tt9xi3nvvPWOMMU2aNDHdunXLs46KFSuayZMnO7XNnDnTBAcHO5abNm1q/vWvfzn1efjhh03btm0dy5LMCy+84Fg+ffq0sdlsZtmyZcYYYzp06GD69OmTZx0Xmzx5sqlYsaJTW0REhKlXr55T27///W9To0YNk5WV5Wh75513TJEiRUxmZqbjdf98bzMyMkxAQIDp0aOHoy0hIcFIMhs3bsyzpn//+9+mevXqxhhjdu3aZYKCgkxGRoYZP3686dq1qzHGmNmzZxtfX1+Tmpqa7/EZMGCA0/ONGzc2t912m2O5dOnSZvz48Y7l8+fPm3LlyuX4/AwePNixXLFiRdO9e3fHclZWlilZsqSZPn26McaY6dOnm5CQEKdj57///a+RZLZv326MubIxy+tYNCb/OZIz3QAAAAAsVbduXafl0qVL69ixY3n237p1q06fPq2QkBAVKVLE8Thw4ID2798vSdqxY4datGhxVXXFxcWpWbNmTm3NmjVzmlTs4voDAgIUGBjoqP/JJ5/U/PnzVa9ePQ0fPlyxsbEFqiX7LP0/a2vSpInT70Q3a9ZMp0+f1uHDh3OtzdPTUyEhIapTp46jrVSpUpJ0yfc7KipKe/fu1ZEjR7RmzRrddddd8vT0VEREhOOy8zVr1ujOO++Uv79/vsYnu/5/+udycnKyEhISnNq8vLxyvA+5+ec+22w2hYWFOfZvz549qlu3rvz8/Bx97rjjDqfXu2rM8ouJ1AAAAABY6uLZp202m7KysvLsn5WVpdKlSzvdZ5wt+55sV02I9s9QK134Sa+L2y5Vf5s2bXTw4EF99dVX+vbbb9WiRQs99dRTevPNN6+ojoCAgMvWYf7vfud/tudW2z/bsvte6v1u1qyZfHx8tGbNGq1evVoRERGSLvwhIDk5WXv37tXq1asd99vnZ3ysdKnxuNT7ls1VY5ZfnOkGAAAAUKjUr19fiYmJ8vLyUtWqVZ0e2ffe1q1bV6tWrcpzHT4+PsrMzLzkdsLDw7VhwwanttjYWIWHX9kEciVKlFDv3r01Z84cTZkyJcfEXgVRq1YtxcbGOgXG2NhYBQYGqmzZsle9/n/y9/dX48aNtWbNGq1bt06RkZGSLpx5btq0qT766CP98ccfjknU8jM+4eHh2rRpk9N2/rkcHBys0qVLO7VlZGRo69atV7UvNWvW1E8//aT09HRH25YtW3L0s2LM8kLoBgAAAFCo3HvvvWrSpIkeeOABffPNN/rjjz8UGxurF154wRGgRo8erXnz5mn06NGKi4vTzz//rAkTJjjWUalSJa1bt05//vmnkpKSct3Oc889p1mzZundd9/Vvn37NGnSJC1atEjDhg3Ld60vvfSSPv/8c/3222/atWuXvvzyyysO7bkZOHCgDh06pEGDBunXX3/V559/rtGjR2vIkCHy8HB9jIuKitL8+fN19uxZ1a9f39EeERGh//znP45gLuVvfAYPHqwPP/xQH374ofbu3avRo0dr165dTtscPHiwxo8fr8WLF+vXX3/VwIEDdfLkyavaj65duyorK0tPPPGE4uLi9M033zjOYGefAbdqzPLC5eUAAAB5iI+Pz/M/1nF9Cg0NtfT3pq+V1NS4y3e6jrZzMZvNpq+//lqjRo1S37599ddffyksLEzNmzd33KMcGRmphQsX6pVXXtH48eMVFBSk5s2bO9YxZswY9e/fX7fccovS09NzXGIsSQ888IDeeustvfHGG3rmmWdUuXJlzZw503GmNz98fHw0cuRI/fHHH/L399fdd9+t+fPnX/V7ULZsWX399dd67rnndNttt6l48eLq16+fXnjhhated26ioqI0ZswY3XffffLy+v8xMSIiQi+88IJatGghX19fSfkbn0ceeUT79+/XiBEjlJaWpgcffFBPPvmkvvnmG8e6hw4dqoSEBPXu3VseHh7q27evOnXqpOTk5ALvR1BQkL744gs9+eSTqlevnurUqaOXXnpJXbt2ddznbdWY5cVmcvv03WROnTql4OBgJScnKygoyN3lAACAQiA+Pl41aoQrLS33n9nB9cnPz649e+IKffBOS0vTgQMHVLlyZacJodzxubxe3jMgL//73//Up08fJScnX/FcAHkdi1L+cyRnugEAAHKRlJSktLRUhYfPkd1u3WWHuHZSU+MUF9ddSUlJ122ArFChgvbsibumV2DcKFcH4Obx0UcfqUqVKipbtqx27typESNGqEuXLi6bfO9KEboBAAAuwW4PV2Bg/ct3BK6RChUqEIKBS0hMTNRLL72kxMRElS5dWg8//LBee+01t9VD6AYAAAAA3DCGDx+u4cOHu7sMB2YvBwAAAADAIoRuAAAAAAAsQugGAAAAAMAihG4AAAAAACxC6AYAAAAAwCKEbgAAAAAALMJPhgEAAADXkfj4eCUlJV2z7YWGht6Qvwveu3dvnTx5UkuWLHF3KS6Tn32KjIxUvXr1NGXKFElSpUqVFB0drejo6GtS482I0A0AAABcJ+Lj4xVeo4ZS09Ku2Tbtfn6K27MnX8HbZrNd8vlevXpp1qxZLqrMWrNmzVJ0dLROnjx5Vet59NFHlZycrGXLljnali1bprZt2+qFF17QK6+84mh/5ZVXNH36dB05cuSqtnklNm/erICAgGu2vZsRoRsAAAC4TiQlJSk1LU1zwsMVbrdbvr241FR1j4tTUlJSvkJ3QkKC4/8vWLBAL730kvbs2eNo8/f3d+p//vx5eXt7u67gQigqKkrDhg1TRkaGvLwuxK81a9aofPnyWr16tVPfNWvWKCoq6prWV6JEiWu6vZsR93QDAAAA15lwu131AwMtf1xpsA8LC3M8goODZbPZHMtpaWkqWrSoPvnkE0VGRsrPz09z5szR33//rccee0zlypWT3W5XnTp1NG/ePKf1ZmVl6fXXX1fVqlXl6+urChUq6LXXXnM8/+eff+qRRx5RsWLFFBISoo4dO+qPP/5wPJ+ZmakhQ4aoaNGiCgkJ0fDhw2WMyXM/1qxZoz59+ig5OVk2m002m00xMTGSpBMnTqhnz54qVqyY7Ha72rRpo3379uW5rqioKJ0+fVpbtmxxWv/zzz+vzZs3KzU1VZJ07tw5bdy40RG6XbFPZ86cUc+ePVWkSBGVLl1aEydOzFFfpUqVHJeaSxeuVvjggw/UqVMn2e12VatWTUuXLnV6zdKlS1WtWjX5+/srKipKs2fPls1mc1wVcPDgQXXo0EHFihVTQECAbr31Vn399dd5vkc3OkI3AAAAgGtmxIgReuaZZxQXF6fWrVsrLS1NDRo00JdffqlffvlFTzzxhHr06KEffvjB8ZqRI0fq9ddf14svvqjdu3dr7ty5KlWqlCQpNTVVUVFRKlKkiNatW6cNGzaoSJEiuu+++3Tu3DlJ0sSJE/Xhhx9qxowZ2rBhg44fP67FixfnWWPTpk01ZcoUBQUFKSEhQQkJCRo2bJikC/dNb9myRUuXLtXGjRtljFHbtm11/vz5XNdVvXp1lSlTxnFWOyUlRdu2bdPDDz+sW265Rd9//70kadOmTTp79qyioqJctk/PPfecVq9ercWLF2vFihVas2aNtm7detkxevnll9WlSxf99NNPatu2rbp166bjx49Lkv744w899NBDeuCBB7Rjxw71799fo0aNcnr9U089pfT0dK1bt04///yzXn/9dRUpUuSy271RcXk5AAAAgGsmOjpanTt3dmrLDrSSNGjQIC1fvlwLFy5U48aNlZKSorfeektvv/22evXqJUm65ZZbdNddd0mS5s+fLw8PD33wwQeOe8pnzpypokWLas2aNWrVqpWmTJmikSNH6sEHH5Qkvfvuu/rmm2/yrNHHx8fpTH22ffv2aenSpfr+++/VtGlTSdL//vc/lS9fXkuWLNHDDz+c6/oiIyO1Zs0ajRw5UuvXr1f16tVVokQJRUREaM2aNWrZsqXjkvNbbrlFH3744VXv0+nTpzVjxgx99NFHatmypSRp9uzZKleu3OWGSL1799Zjjz0mSRo7dqymTp2qH3/8Uffdd5/effdd1ahRQ2+88YYkqUaNGvrll1+crjyIj4/Xgw8+qDp16kiSqlSpctlt3sgI3QAAAACumYYNGzotZ2Zmavz48VqwYIH+/PNPpaenKz093TG5V1xcnNLT09WiRYtc17d161b99ttvCgwMdGpPS0vT/v37lZycrISEBDVp0sTxnJeXlxo2bHjJS8xzExcXJy8vLzVu3NjRFhISoho1aiguLi7P10VFRSk6Olrnz5/XmjVrFBkZKUmKiIjQ1KlTJV245Pyee+5x2T7t379f586dc+pTvHhx1ahR47L7WbduXcf/DwgIUGBgoI4dOyZJ2rNnjxo1auTU/4477nBafuaZZ/Tkk09qxYoVuvfee/Xggw86rfNmQ+gGAAAAcM1cPFP2xIkTNXnyZE2ZMkV16tRRQECAoqOjHZdRXzz52sWysrLUoEED/e9//8vxnKsnCcsrpBtjLjlze1RUlM6cOaPNmzdr9erVeu655yRdCN09e/bU8ePHtXHjRseZfFfs05X+QeGfLp7czmazKSsry7Hei/f14m09/vjjat26tb766iutWLFC48aN08SJEzVo0KAC13Q9455uAAAAAG6zfv16dezYUd27d9dtt92mKlWqOE1Mlj1h16pVq3J9ff369bVv3z6VLFlSVatWdXoEBwcrODhYpUuX1qZNmxyvycjIuOy9zT4+PsrMzHRqq1WrljIyMpzuN//777+1d+9ehYeH57muW265ReXLl9fSpUu1Y8cORURESJJKly6tSpUqaeLEiUpLS3NMouaKfapataq8vb2d+pw4cUJ79+695H5fTs2aNbV582antn9OEpetfPnyGjBggBYtWqShQ4fqv//971Vt93pG6AYAAADgNlWrVtXKlSsVGxuruLg49e/fX4mJiY7n/fz8NGLECA0fPlwfffSR9u/fr02bNmnGjBmSpG7duik0NFQdO3bU+vXrdeDAAa1du1aDBw/W4cOHJUmDBw/W+PHjtXjxYv36668aOHDgZX9/u1KlSjp9+rRWrVp14afaUlNVrVo1dezYUf/617+0YcMG7dy5U927d1fZsmXVsWPHS64vKipK06ZNU9WqVR2TwEn//xLzKlWqOH6WzRX7VKRIEfXr10/PPfecVq1apV9++UW9e/eWh8fVRcD+/fvr119/1YgRI7R371598sknjt9ezz4DHh0drW+++UYHDhzQtm3b9N13313yjxI3Oi4vBwAAAK4zcf/3M1M3wnZefPFFHThwQK1bt5bdbtcTTzyhBx54QMnJyU59vLy89NJLL+nIkSMqXbq0BgwYIEmy2+1at26dRowYoc6dOyslJUVly5ZVixYtFBQUJEkaOnSoEhISHKGzb9++6tSpk9M2Lta0aVMNGDBAjzzyiP7++2+NHj1aMTExmjlzpgYPHqz27dvr3Llzat68ub7++uvL/t54VFSUPvroI8f93NkiIiL0wQcfqEuXLo42V+3TG2+8odOnT+v+++9XYGCghg4desl9zo/KlSvr008/1dChQ/XWW2+pSZMmGjVqlJ588kn5+vpKunCf/lNPPaXDhw8rKChI9913nyZPnnxV272e2czVXOx/gzh16pSCg4OVnJzs+BADAICb27Zt29SgQQM1aLBVgYH13V0OXCAlZZu2bm2grVu3qn79wj2maWlpOnDggCpXriw/Pz9He3x8vMJr1FBqWto1q8Xu56e4PXscZ2GBi7322mt69913dejQIXeX4nJ5HYtS/nMkZ7oBAACA60SFChUUt2ePkpKSrtk2Q0NDCdxwMm3aNDVq1EghISH6/vvv9cYbb+jpp592d1mFFqEbAAAAuI5UqFCBEAy32rdvn1599VUdP35cFSpU0NChQzVy5Eh3l1VoEboBAAAAAPk2efLkm/oe7StF6AYAAPg/8fHxjst24+LiJEmpqXHuLAku4u0d6u4SANykCN0AAADKe4KquLjubqoIruTl4afwWz91dxkAbkKEbgAAAOnC7/CmpWlOeLjC7XadSU3V7rg4BfiHy8PT7u7ycBUOZKXqhdQ4ZWScdHcpVywrK8vdJQA3NVccg4RuAACAfwi321U/MFApkjIkBXra5ekV6O6ycDUy3F3AlfPx8ZGHh4eOHDmiEiVKyMfHRzabzd1lATcNY4zOnTunv/76Sx4eHvLx8SnwugjdAAAAQCHj4eGhypUrKyEhQUeOHHF3OcBNy263q0KFCvLw8CjwOgjdAAAAQCHk4+OjChUqKCMjQ5mZme4uB7jpeHp6ysvL66qvMiF0AwAAAIWUzWaTt7e3vL293V0KgAIq+DlyAAAAAABwSYRuAAAAAAAsQugGAAAAAMAihG4AAAAAACxC6AYAAAAAwCKEbgAAAAAALELoBgAAAADAIoRuAAAAAAAsQugGAAAAAMAihG4AAAAAACxC6AYAAAAAwCKEbgAAAAAALELoBgAAAADAIoUmdI8bN042m03R0dGONmOMYmJiVKZMGfn7+ysyMlK7du1yel16eroGDRqk0NBQBQQE6P7779fhw4evcfUAAAAAAORUKEL35s2b9f7776tu3bpO7RMmTNCkSZP09ttva/PmzQoLC1PLli2VkpLi6BMdHa3Fixdr/vz52rBhg06fPq327dsrMzPzWu8GAAAAAABO3B66T58+rW7duum///2vihUr5mg3xmjKlCkaNWqUOnfurNq1a2v27NlKTU3V3LlzJUnJycmaMWOGJk6cqHvvvVe333675syZo59//lnffvutu3YJAAAAAABJhSB0P/XUU2rXrp3uvfdep/YDBw4oMTFRrVq1crT5+voqIiJCsbGxkqStW7fq/PnzTn3KlCmj2rVrO/rkJj09XadOnXJ6AAAAAADgal7u3Pj8+fO1bds2bd68OcdziYmJkqRSpUo5tZcqVUoHDx509PHx8XE6Q57dJ/v1uRk3bpxefvnlqy0fAAAAAIBLctuZ7kOHDmnw4MGaM2eO/Pz88uxns9mclo0xOdoudrk+I0eOVHJysuNx6NChKyseAAAAAIB8cFvo3rp1q44dO6YGDRrIy8tLXl5eWrt2rf7zn//Iy8vLcYb74jPWx44dczwXFhamc+fO6cSJE3n2yY2vr6+CgoKcHgAAAAAAuJrbQneLFi30888/a8eOHY5Hw4YN1a1bN+3YsUNVqlRRWFiYVq5c6XjNuXPntHbtWjVt2lSS1KBBA3l7ezv1SUhI0C+//OLoAwAAAACAu7jtnu7AwEDVrl3bqS0gIEAhISGO9ujoaI0dO1bVqlVTtWrVNHbsWNntdnXt2lWSFBwcrH79+mno0KEKCQlR8eLFNWzYMNWpUyfHxGwAAAAAAFxrbp1I7XKGDx+us2fPauDAgTpx4oQaN26sFStWKDAw0NFn8uTJ8vLyUpcuXXT27Fm1aNFCs2bNkqenpxsrBwAAAACgkIXuNWvWOC3bbDbFxMQoJiYmz9f4+flp6tSpmjp1qrXFAQAAAABwhdz+O90AAAAAANyoCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBF3Bq6p0+frrp16yooKEhBQUFq0qSJli1b5njeGKOYmBiVKVNG/v7+ioyM1K5du5zWkZ6erkGDBik0NFQBAQG6//77dfjw4Wu9KwAAAAAA5ODW0F2uXDmNHz9eW7Zs0ZYtW3TPPfeoY8eOjmA9YcIETZo0SW+//bY2b96ssLAwtWzZUikpKY51REdHa/HixZo/f742bNig06dPq3379srMzHTXbgEAAAAAIMnNobtDhw5q27atqlevrurVq+u1115TkSJFtGnTJhljNGXKFI0aNUqdO3dW7dq1NXv2bKWmpmru3LmSpOTkZM2YMUMTJ07Uvffeq9tvv11z5szRzz//rG+//daduwYAAAAAQOG5pzszM1Pz58/XmTNn1KRJEx04cECJiYlq1aqVo4+vr68iIiIUGxsrSdq6davOnz/v1KdMmTKqXbu2ow8AAAAAAO7i5e4Cfv75ZzVp0kRpaWkqUqSIFi9erFq1ajlCc6lSpZz6lypVSgcPHpQkJSYmysfHR8WKFcvRJzExMc9tpqenKz093bF86tQpV+0OAAAAAAAObj/TXaNGDe3YsUObNm3Sk08+qV69emn37t2O5202m1N/Y0yOtotdrs+4ceMUHBzseJQvX/7qdgIAAAAAgFy4PXT7+PioatWqatiwocaNG6fbbrtNb731lsLCwiQpxxnrY8eOOc5+h4WF6dy5czpx4kSefXIzcuRIJScnOx6HDh1y8V4BAAAAAFAIQvfFjDFKT09X5cqVFRYWppUrVzqeO3funNauXaumTZtKkho0aCBvb2+nPgkJCfrll18cfXLj6+vr+Jmy7AcAAAAAAK7m1nu6//3vf6tNmzYqX768UlJSNH/+fK1Zs0bLly+XzWZTdHS0xo4dq2rVqqlatWoaO3as7Ha7unbtKkkKDg5Wv379NHToUIWEhKh48eIaNmyY6tSpo3vvvdeduwYAAAAAgHtD99GjR9WjRw8lJCQoODhYdevW1fLly9WyZUtJ0vDhw3X27FkNHDhQJ06cUOPGjbVixQoFBgY61jF58mR5eXmpS5cuOnv2rFq0aKFZs2bJ09PTXbsFAAAAAIAkN4fuGTNmXPJ5m82mmJgYxcTE5NnHz89PU6dO1dSpU11cHQAAAAAAV6fQ3dMNAAAAAMCNwu2/0w0AwI0gPj5eSUlJ7i4DVyEuLk6SdCY1VSn/978AAFwtQjcAAFcpPj5eNWqEKy2NkHYj2B0Xp4x/LGcZI2aKAQAUFKEbAICrlJSUpLS0VIWHz5HdHu7uclBAqalxiovrrgD/cAV62nU+47jS0g7IGOPu0gAA1zFCNwAALmK3hyswsL67y8BV8vC0y9MrUJlZXLkAALh6TKQGAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUKFLoPHDjg6joAAAAAALjhFCh0V61aVVFRUZozZ47S0tJcXRMAAAAAADeEAoXunTt36vbbb9fQoUMVFham/v3768cff3R1bQAAAAAAXNcKFLpr166tSZMm6c8//9TMmTOVmJiou+66S7feeqsmTZqkv/76y9V1AgAAAABw3bmqidS8vLzUqVMnffLJJ3r99de1f/9+DRs2TOXKlVPPnj2VkJDgqjoBAAAAALjuXFXo3rJliwYOHKjSpUtr0qRJGjZsmPbv36/vvvtOf/75pzp27OiqOgEAAAAAuO54FeRFkyZN0syZM7Vnzx61bdtWH330kdq2bSsPjwsZvnLlynrvvfdUs2ZNlxYLAAAAAMD1pEChe/r06erbt6/69OmjsLCwXPtUqFBBM2bMuKriAAAAAAC4nhUodO/bt++yfXx8fNSrV6+CrB4AAAAAgBtCge7pnjlzphYuXJijfeHChZo9e/ZVFwUAAAAAwI2gQKF7/PjxCg0NzdFesmRJjR079qqLAgAAAADgRlCg0H3w4EFVrlw5R3vFihUVHx9/1UUBAAAAAHAjKNA93SVLltRPP/2kSpUqObXv3LlTISEhrqgLAIBCLT4+XklJSZKkuLg4SVJqapw7S8JVYvwAAFYoUOh+9NFH9cwzzygwMFDNmzeXJK1du1aDBw/Wo48+6tICAQAobOLj4xVeo4ZS09Kc2uPiurupIriSMefcXQIA4AZSoND96quv6uDBg2rRooW8vC6sIisrSz179uSebgDADS8pKUmpaWmaEx6ucLtdZ1JTtTsuTgH+4fLwtLu7PBTQ9xl/a1raHzJZGe4uBQBwAylQ6Pbx8dGCBQv0yiuvaOfOnfL391edOnVUsWJFV9cHAEChFW63q35goFIkZUgK9LTL0yvQ3WWhgA5kprq7BADADahAoTtb9erVVb16dVfVAgAAAADADaVAoTszM1OzZs3SqlWrdOzYMWVlZTk9/91337mkOAAAAAAArmcFCt2DBw/WrFmz1K5dO9WuXVs2m83VdQEAAAAAcN0rUOieP3++PvnkE7Vt29bV9QAAAAAAcMPwKMiLfHx8VLVqVVfXAgAAAADADaVAoXvo0KF66623ZIxxdT0AAAAAANwwCnR5+YYNG7R69WotW7ZMt956q7y9vZ2eX7RokUuKAwAAAADgelag0F20aFF16tTJ1bUAAAAAAHBDKVDonjlzpqvrAAAAAADghlOge7olKSMjQ99++63ee+89paSkSJKOHDmi06dPu6w4AAAAAACuZwU6033w4EHdd999io+PV3p6ulq2bKnAwEBNmDBBaWlpevfdd11dJwAAAAAA150CnekePHiwGjZsqBMnTsjf39/R3qlTJ61atcplxQEAAAAAcD0r8Ozl33//vXx8fJzaK1asqD///NMlhQEAAAAAcL0r0JnurKwsZWZm5mg/fPiwAgMDr7ooAAAAAABuBAUK3S1bttSUKVMcyzabTadPn9bo0aPVtm1bV9UGAAAAAMB1rUCXl0+ePFlRUVGqVauW0tLS1LVrV+3bt0+hoaGaN2+eq2sEAAAAAOC6VKDQXaZMGe3YsUPz5s3Ttm3blJWVpX79+qlbt25OE6sBAAAAAHAzK1DoliR/f3/17dtXffv2dWU9AAAAAADcMAoUuj/66KNLPt+zZ88CFQMAAAAAwI2kQKF78ODBTsvnz59XamqqfHx8ZLfbCd0AAAAAAKiAs5efOHHC6XH69Gnt2bNHd911FxOpAQAAAADwfwoUunNTrVo1jR8/PsdZcAAAAAAAblYuC92S5OnpqSNHjrhylQAAAAAAXLcKdE/30qVLnZaNMUpISNDbb7+tZs2auaQwAAAAAACudwUK3Q888IDTss1mU4kSJXTPPfdo4sSJrqgLAAAAAIDrXoFCd1ZWlqvrAAAAAADghuPSe7oBAAAAAMD/V6Az3UOGDMl330mTJhVkEwAAAAAAXPcKFLq3b9+ubdu2KSMjQzVq1JAk7d27V56enqpfv76jn81mc02VAAAAAABchwoUujt06KDAwEDNnj1bxYoVkySdOHFCffr00d13362hQ4e6tEgAAAAAAK5HBbqne+LEiRo3bpwjcEtSsWLF9OqrrzJ7OQAAAAAA/6dAofvUqVM6evRojvZjx44pJSXlqosCAAAAAOBGUKDQ3alTJ/Xp00effvqpDh8+rMOHD+vTTz9Vv3791LlzZ1fXCAAAAADAdalA93S/++67GjZsmLp3767z589fWJGXl/r166c33njDpQUCAAAAAHC9KlDottvtmjZtmt544w3t379fxhhVrVpVAQEBrq4PAAAAAIDrVoEuL8+WkJCghIQEVa9eXQEBATLGuKouAAAAAACuewUK3X///bdatGih6tWrq23btkpISJAkPf744/xcGAAAAAAA/6dAofvZZ5+Vt7e34uPjZbfbHe2PPPKIli9f7rLiAAAAAAC4nhXonu4VK1bom2++Ubly5Zzaq1WrpoMHD7qkMAAAAAAArncFOtN95swZpzPc2ZKSkuTr63vVRQEAAAAAcCMoUOhu3ry5PvroI8eyzWZTVlaW3njjDUVFRbmsOAAAAAAArmcFurz8jTfeUGRkpLZs2aJz585p+PDh2rVrl44fP67vv//e1TUCAAAAAHBdKtCZ7lq1aumnn37SHXfcoZYtW+rMmTPq3Lmztm/frltuucXVNQIAAAAAcF264jPd58+fV6tWrfTee+/p5ZdftqImAAAAAABuCFd8ptvb21u//PKLbDabFfUAAAAAAHDDKNDl5T179tSMGTNcXQsAAAAAADeUAk2kdu7cOX3wwQdauXKlGjZsqICAAKfnJ02a5JLiAAAAAAC4nl1R6P79999VqVIl/fLLL6pfv74kae/evU59uOwcAAAAAIALrih0V6tWTQkJCVq9erUk6ZFHHtF//vMflSpVypLiAAAAAAC4nl3RPd3GGKflZcuW6cyZMwXe+Lhx49SoUSMFBgaqZMmSeuCBB7Rnz54c24yJiVGZMmXk7++vyMhI7dq1y6lPenq6Bg0apNDQUAUEBOj+++/X4cOHC1wXAAAAAACuUKCJ1LJdHMKv1Nq1a/XUU09p06ZNWrlypTIyMtSqVSunID9hwgRNmjRJb7/9tjZv3qywsDC1bNlSKSkpjj7R0dFavHix5s+frw0bNuj06dNq3769MjMzr6o+AAAAAACuxhVdXm6z2XLcs30193AvX77caXnmzJkqWbKktm7dqubNm8sYoylTpmjUqFHq3LmzJGn27NkqVaqU5s6dq/79+ys5OVkzZszQxx9/rHvvvVeSNGfOHJUvX17ffvutWrduXeD6AAAAAAC4GlcUuo0x6t27t3x9fSVJaWlpGjBgQI7ZyxctWlSgYpKTkyVJxYsXlyQdOHBAiYmJatWqlaOPr6+vIiIiFBsbq/79+2vr1q06f/68U58yZcqodu3aio2NJXQDAAAAANzmikJ3r169nJa7d+/uskKMMRoyZIjuuusu1a5dW5KUmJgoSTkmaitVqpQOHjzo6OPj46NixYrl6JP9+oulp6crPT3dsXzq1CmX7QcAAAAAANmuKHTPnDnTqjr09NNP66efftKGDRtyPHfxJezGmMte1n6pPuPGjdPLL79c8GIBAAAAAMiHq5pIzVUGDRqkpUuXavXq1SpXrpyjPSwsTJJynLE+duyY4+x3WFiYzp07pxMnTuTZ52IjR45UcnKy43Ho0CFX7g4AAAAAAJLcHLqNMXr66ae1aNEifffdd6pcubLT85UrV1ZYWJhWrlzpaDt37pzWrl2rpk2bSpIaNGggb29vpz4JCQn65ZdfHH0u5uvrq6CgIKcHAAAAAACudkWXl7vaU089pblz5+rzzz9XYGCg44x2cHCw/P39ZbPZFB0drbFjx6patWqqVq2axo4dK7vdrq5duzr69uvXT0OHDlVISIiKFy+uYcOGqU6dOo7ZzAEAAAAAcAe3hu7p06dLkiIjI53aZ86cqd69e0uShg8frrNnz2rgwIE6ceKEGjdurBUrVigwMNDRf/LkyfLy8lKXLl109uxZtWjRQrNmzZKnp+e12hUAAAAAAHJwa+g2xly2j81mU0xMjGJiYvLs4+fnp6lTp2rq1KkurA4AAAAAgKtTKCZSAwAAAADgRkToBgAAAADAIoRuAAAAAAAsQugGAAAAAMAihG4AAAAAACxC6AYAAAAAwCKEbgAAAAAALELoBgAAAADAIoRuAAAAAAAsQugGAAAAAMAihG4AAAAAACxC6AYAAAAAwCKEbgAAAAAALELoBgAAAADAIoRuAAAAAAAsQugGAAAAAMAihG4AAAAAACxC6AYAAAAAwCKEbgAAAAAALELoBgAAAADAIoRuAAAAAAAsQugGAAAAAMAihG4AAAAAACxC6AYAAAAAwCKEbgAAAAAALOLl7gIAAACAaykuLs7dJcDFQkNDVaFCBXeXAeSK0A0AAICbwvnzSZI81L17d3eXAhfz87Nrz544gjcKJUI3AAAAbgoZGSmSslSp0n8VElLf3eXARVJT4xQX111JSUmEbhRKhG4AAADcVPz9aygwkNAN4NpgIjUAAAAAACzCmW4AAADcFNLTj0iSzp7do5SUADdXA1c5dy7B3SUAl0ToBgAAwA0tKeucPCQlJEyXJP3xx7/0xx9uLQku5GnzdXcJwCURugEAAHBDSzEZypI02ruESp//S35+NeTtVcTdZcEFDmSl6oVUfgIOhRuhGwAAADeFSjZvVZRk97DLxyvQ3eXAFTLcXQBweUykBgAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFjEraF73bp16tChg8qUKSObzaYlS5Y4PW+MUUxMjMqUKSN/f39FRkZq165dTn3S09M1aNAghYaGKiAgQPfff78OHz58DfcCAAAAAIDcuTV0nzlzRrfddpvefvvtXJ+fMGGCJk2apLffflubN29WWFiYWrZsqZSUFEef6OhoLV68WPPnz9eGDRt0+vRptW/fXpmZmddqNwAAAAAAyJWXOzfepk0btWnTJtfnjDGaMmWKRo0apc6dO0uSZs+erVKlSmnu3Lnq37+/kpOTNWPGDH388ce69957JUlz5sxR+fLl9e2336p169bXbF8AAAAAALhYob2n+8CBA0pMTFSrVq0cbb6+voqIiFBsbKwkaevWrTp//rxTnzJlyqh27dqOPrlJT0/XqVOnnB4AAAAAALhaoQ3diYmJkqRSpUo5tZcqVcrxXGJionx8fFSsWLE8++Rm3LhxCg4OdjzKly/v4uoBAAAAACjEoTubzWZzWjbG5Gi72OX6jBw5UsnJyY7HoUOHXFIrAAAAAAD/VGhDd1hYmCTlOGN97Ngxx9nvsLAwnTt3TidOnMizT258fX0VFBTk9AAAAAAAwNUKbeiuXLmywsLCtHLlSkfbuXPntHbtWjVt2lSS1KBBA3l7ezv1SUhI0C+//OLoAwAAAACAu7h19vLTp0/rt99+cywfOHBAO3bsUPHixVWhQgVFR0dr7NixqlatmqpVq6axY8fKbrera9eukqTg4GD169dPQ4cOVUhIiIoXL65hw4apTp06jtnMbyTx8fFKSkpydxmwQGhoqCpUqODuMgAAAAC4mFtD95YtWxQVFeVYHjJkiCSpV69emjVrloYPH66zZ89q4MCBOnHihBo3bqwVK1YoMDDQ8ZrJkyfLy8tLXbp00dmzZ9WiRQvNmjVLnp6e13x/rBQfH6/wGjWUmpbm7lJgAbufn+L27CF4AwAAADcYt4buyMhIGWPyfN5msykmJkYxMTF59vHz89PUqVM1depUCyosPJKSkpSalqY54eEKt9vdXQ5cKC41Vd3j4pSUlEToBgAAAG4wbg3duHLhdrvq/+NMPwAAAACg8Cq0E6kBAAAAAHC9I3QDAAAAAGARQjcAAAAAABYhdAMAAAAAYBFCNwAAAAAAFiF0AwAAAABgEUI3AAAAAAAWIXQDAAAAAGARQjcAAAAAABYhdAMAAAAAYBFCNwAAAAAAFiF0AwAAAABgEUI3AAAAAAAWIXQDAAAAAGARQjcAAAAAABbxcncBAHAzio+PV1JSkrvLQAHFxcVJks6kpirl//4XAAAgN4RuALjG4uPjVaNGuNLSCGrXu91xccr4x3KWMfJ0WzUAAKAwInQDwDWWlJSktLRUhYfPkd0e7u5yUACpqXGKi+uuAP9wBXradT7juNLSDsgY4+7SAABAIUPoBgA3sdvDFRhY391l4Cp4eNrl6RWozCyuWgAAALljIjUAAAAAACxC6AYAAAAAwCKEbgAAAAAALELoBgAAAADAIoRuAAAAAAAsQugGAAAAAMAihG4AAAAAACxC6AYAAAAAwCKEbgAAAAAALELoBgAAAADAIoRuAAAAAAAsQugGAAAAAMAihG4AAAAAACxC6AYAAAAAwCKEbgAAAAAALELoBgAAAADAIoRuAAAAAAAsQugGAAAAAMAihG4AAAAAACxC6AYAAAAAwCKEbgAAAAAALELoBgAAAADAIoRuAAAAAAAsQugGAAAAAMAihG4AAAAAACzi5e4CAOBGFh8fr6SkJKe2uLg4SVJqapw7SoILMHYAACC/CN0AYJH4+HiF16ih1LS0XJ+Pi+t+jSuCqxlzzt0lAACAQo7QDQAWSUpKUmpamuaEhyvcbne0n0lN1e64OAX4h8vD036JNaCw+j7jb01L+0MmK8PdpQAAgEKO0A0AFgu321U/MNCxnCIpQ1Kgp12eXoF5vg6F14HMVHeXAAAArhNMpAYAAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWITQDQAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWMTL3QUAAAAAwNX6+uuvFRcX5+4y4CJVqlRRkyZN3F2GSxC6AQAAAFyXkrLOyUNSlqQXX3zR3eXAhTwkbYiNvSGCN6EbAAAAwHUpxWQoS9K/JVXzqSgvT7u7S4IL7M9MVsy5I/r9998J3QAAAADgbhUkhXsVl49PsLtLgSukS9IRd1fhMkykBgAAAACARQjdAAAAAABYhNANAAAAAIBFCN0AAAAAAFiE0A0AAAAAgEUI3QAAAAAAWOSGCd3Tpk1T5cqV5efnpwYNGmj9+vXuLgkAAAAAcJO7IUL3ggULFB0drVGjRmn79u26++671aZNG8XHx7u7NAAAAADATczL3QW4wqRJk9SvXz89/vjjkqQpU6bom2++0fTp0zVu3Dg3Vwdcnfj4eCUlJbm7DBRAXFycJOlMaqpS/tF+JjXVPQUBAADgmrvuQ/e5c+e0detWPf/8807trVq1UmxsrJuqAlwjPj5eNWqEKy2NkHY92x0Xp4xc2rOMkec1rwYAAADX0nUfupOSkpSZmalSpUo5tZcqVUqJiYm5viY9PV3p6emO5eTkZEnSqVOnrCv0Kp0+fVqStDUlRaczM91cDVxpz/+d9dy6datjnB3P7dmjtLRUlSzZQz4+Ye4oD1fh3LlEHTv2sf7wKiXj4eNoz8w8o/OZx+V77i95ZZ6+xBpQWP2eceHfjV0ZJ3TWZCoj85TSJfmeT5JX1hn3FocCY1xvXNlj+2vGGZ0UY3ojyR7bfZKyGNcbxu+ZF3JZampqoc5o2bUZYy7Zz2Yu16OQO3LkiMqWLavY2Fg1adLE0f7aa6/p448/1q+//prjNTExMXr55ZevZZkAAAAAgBvQoUOHVK5cuTyfv+7PdIeGhsrT0zPHWe1jx47lOPudbeTIkRoyZIhjOSsrS8ePH1dISIhsNpul9RbUqVOnVL58eR06dEhBQUHuLgfXCON+c2Lcb06M+82Jcb85Me43J8b9xmOMUUpKisqUKXPJftd96Pbx8VGDBg20cuVKderUydG+cuVKdezYMdfX+Pr6ytfX16mtaNGiVpbpMkFBQRykNyHG/ebEuN+cGPebE+N+c2Lcb06M+40lODj4sn2u+9AtSUOGDFGPHj3UsGFDNWnSRO+//77i4+M1YMAAd5cGAAAAALiJ3RCh+5FHHtHff/+tMWPGKCEhQbVr19bXX3+tihUrurs0AAAAAMBN7IYI3ZI0cOBADRw40N1lWMbX11ejR4/OcVk8bmyM+82Jcb85Me43J8b95sS435wY95vXdT97OQAAAAAAhZWHuwsAAAAAAOBGRegGAAAAAMAihG4AAAAAACxC6HajadOmqXLlyvLz81ODBg20fv36S/Zfu3atGjRoID8/P1WpUkXvvvtujj6fffaZatWqJV9fX9WqVUuLFy+2qnwUkKvHfdasWbLZbDkeaWlpVu4GrtCVjHtCQoK6du2qGjVqyMPDQ9HR0bn243gv/Fw97hzvhd+VjPmiRYvUsmVLlShRQkFBQWrSpIm++eabHP041gs/V487x/r14UrGfcOGDWrWrJlCQkLk7++vmjVravLkyTn6cbzfmAjdbrJgwQJFR0dr1KhR2r59u+6++261adNG8fHxufY/cOCA2rZtq7vvvlvbt2/Xv//9bz3zzDP67LPPHH02btyoRx55RD169NDOnTvVo0cPdenSRT/88MO12i1chhXjLklBQUFKSEhwevj5+V2LXUI+XOm4p6enq0SJEho1apRuu+22XPtwvBd+Voy7xPFemF3pmK9bt04tW7bU119/ra1btyoqKkodOnTQ9u3bHX041gs/K8Zd4lgv7K503AMCAvT0009r3bp1iouL0wsvvKAXXnhB77//vqMPx/sNzMAt7rjjDjNgwACntpo1a5rnn38+1/7Dhw83NWvWdGrr37+/ufPOOx3LXbp0Mffdd59Tn9atW5tHH33URVXjalkx7jNnzjTBwcEurxWuc6Xj/k8RERFm8ODBOdo53gs/K8ad471wu5oxz1arVi3z8ssvO5Y51gs/K8adY73wc8W4d+rUyXTv3t2xzPF+4+JMtxucO3dOW7duVatWrZzaW7VqpdjY2Fxfs3Hjxhz9W7durS1btuj8+fOX7JPXOnFtWTXuknT69GlVrFhR5cqVU/v27XP8tRzuU5Bxzw+O98LNqnGXON4LK1eMeVZWllJSUlS8eHFHG8d64WbVuEsc64WZK8Z9+/btio2NVUREhKON4/3GReh2g6SkJGVmZqpUqVJO7aVKlVJiYmKur0lMTMy1f0ZGhpKSki7ZJ6914tqyatxr1qypWbNmaenSpZo3b578/PzUrFkz7du3z5odwRUpyLjnB8d74WbVuHO8F16uGPOJEyfqzJkz6tKli6ONY71ws2rcOdYLt6sZ93LlysnX11cNGzbUU089pccff9zxHMf7jcvL3QXczGw2m9OyMSZH2+X6X9x+pevEtefqcb/zzjt15513Op5v1qyZ6tevr6lTp+o///mPq8rGVbLi2OR4L/xcPUYc74VfQcd83rx5iomJ0eeff66SJUu6ZJ24dlw97hzr14eCjPv69et1+vRpbdq0Sc8//7yqVq2qxx577KrWicKP0O0GoaGh8vT0zPFXq2PHjuX461a2sLCwXPt7eXkpJCTkkn3yWieuLavG/WIeHh5q1KgRfw0vJAoy7vnB8V64WTXuF+N4LzyuZswXLFigfv36aeHChbr33nudnuNYL9ysGveLcawXLlcz7pUrV5Yk1alTR0ePHlVMTIwjdHO837i4vNwNfHx81KBBA61cudKpfeXKlWratGmur2nSpEmO/itWrFDDhg3l7e19yT55rRPXllXjfjFjjHbs2KHSpUu7pnBclYKMe35wvBduVo37xTjeC4+Cjvm8efPUu3dvzZ07V+3atcvxPMd64WbVuF+MY71wcdV3vDFG6enpjmWO9xvYNZ+6DcYYY+bPn2+8vb3NjBkzzO7du010dLQJCAgwf/zxhzHGmOeff9706NHD0f/33383drvdPPvss2b37t1mxowZxtvb23z66aeOPt9//73x9PQ048ePN3FxcWb8+PHGy8vLbNq06ZrvH3JnxbjHxMSY5cuXm/3795vt27ebPn36GC8vL/PDDz9c8/1D7q503I0xZvv27Wb79u2mQYMGpmvXrmb79u1m165djuc53gs/K8ad471wu9Ixnzt3rvHy8jLvvPOOSUhIcDxOnjzp6MOxXvhZMe4c64XflY7722+/bZYuXWr27t1r9u7daz788EMTFBRkRo0a5ejD8X7jInS70TvvvGMqVqxofHx8TP369c3atWsdz/Xq1ctEREQ49V+zZo25/fbbjY+Pj6lUqZKZPn16jnUuXLjQ1KhRw3h7e5uaNWuazz77zOrdwBVy9bhHR0ebChUqGB8fH1OiRAnTqlUrExsbey12BVfgSsddUo5HxYoVnfpwvBd+rh53jvfC70rGPCIiItcx79Wrl9M6OdYLP1ePO8f69eFKxv0///mPufXWW43dbjdBQUHm9ttvN9OmTTOZmZlO6+R4vzHZjPm/WZkAAAAAAIBLcU83AAAAAAAWIXQDAAAAAGARQjcAAAAAABYhdAMAAAAAYBFCNwAAAAAAFiF0AwAAAABgEUI3AAAAAAAWIXQDAAAAAGARQjcAADeBSpUqacqUKe4uQ6mpqXrwwQcVFBQkm82mkydPurskAAAsRegGACAPvXv3ls1mk81mk7e3t6pUqaJhw4bpzJkz7i4tT7NmzVLRokVztG/evFlPPPHEtS/oIrNnz9b69esVGxurhIQEBQcHu7skAAAs5eXuAgAAKMzuu+8+zZw5U+fPn9f69ev1+OOP68yZM5o+fXqOvufPn5e3t7cbqvz/289LiRIlrmEledu/f7/Cw8NVu3Ztd5cCAMA1wZluAAAuwdfXV2FhYSpfvry6du2qbt26acmSJZKkmJgY1atXTx9++KGqVKkiX19fGWMUHx+vjh07qkiRIgoKClKXLl109OhRxzqzX/fee++pfPnystvtevjhh50utc7KytKYMWNUrlw5+fr6ql69elq+fLnj+T/++EM2m02ffPKJIiMj5efnpzlz5qhPnz5KTk52nKGPiYmRlPPy8vzW+PHHH6tSpUoKDg7Wo48+qpSUlEu+X5999pluvfVW+fr6qlKlSpo4caLjucjISE2cOFHr1q2TzWZTZGRknutZunSpGjZsKD8/P4WGhqpz586O5+bMmaOGDRsqMDBQYWFh6tq1q44dO+Z4/sSJE+rWrZtKlCghf39/VatWTTNnznQ8/+eff+qRRx5RsWLFFBISoo4dO+qPP/5wPL9mzRrdcccdCggIUNGiRdWsWTMdPHjwkvsNAEBeCN0AAFwBf39/pzPKv/32mz755BN99tln2rFjhyTpgQce0PHjx7V27VqtXLlS+/fv1yOPPOK0nuzXffHFF1q+fLl27Nihp556yvH8W2+9pYkTJ+rNN9/UTz/9pNatW+v+++/Xvn37nNYzYsQIPfPMM4qLi1OLFi00ZcoUBQUFKSEhQQkJCRo2bFiOfTDG5KvG/fv3a8mSJfryyy/15Zdfau3atRo/fnye783WrVvVpUsXPfroo/r5558VExOjF198UbNmzZIkLVq0SP/617/UpEkTJSQkaNGiRbmu56uvvlLnzp3Vrl07bd++XatWrVLDhg0dz587d06vvPKKdu7cqSVLlujAgQPq3bu34/kXX3xRu3fv1rJlyxQXF6fp06crNDRU0oV7yqOiolSkSBGtW7dOGzZsUJEiRXTffffp3LlzysjI0AMPPKCIiAj99NNP2rhxo5544gnZbLY89xsAgEsyAAAgV7169TIdO3Z0LP/www8mJCTEdOnSxRhjzOjRo423t7c5duyYo8+KFSuMp6eniY+Pd7Tt2rXLSDI//vij43Wenp7m0KFDjj7Lli0zHh4eJiEhwRhjTJkyZcxrr73mVE+jRo3MwIEDjTHGHDhwwEgyU6ZMceozc+ZMExwcnGNfKlasaCZPnnxFNdrtdnPq1ClHn+eee840btw4z/era9eupmXLlk5tzz33nKlVq5ZjefDgwSYiIiLPdRhjTJMmTUy3bt0u2eeffvzxRyPJpKSkGGOM6dChg+nTp0+ufWfMmGFq1KhhsrKyHG3p6enG39/ffPPNN+bvv/82ksyaNWvyvX0AAC6FM90AAFzCl19+qSJFisjPz09NmjRR8+bNNXXqVMfzFStWdLpfOi4uTuXLl1f58uUdbbVq1VLRokUVFxfnaKtQoYLKlSvnWG7SpImysrK0Z88enTp1SkeOHFGzZs2camnWrJnTOiQ5nQHOr/zWWKlSJQUGBjqWS5cu7XQZd27rza3mffv2KTMzM9/17dixQy1atMjz+e3bt6tjx46qWLGiAgMDHZepx8fHS5KefPJJzZ8/X/Xq1dPw4cMVGxvreO3WrVv122+/KTAwUEWKFFGRIkVUvHhxpaWlaf/+/SpevLh69+6t1q1bq0OHDnrrrbeUkJCQ79oBALgYoRsAgEuIiorSjh07tGfPHqWlpWnRokUqWbKk4/mAgACn/saYXC9Fzqs9W/Zz/+xzcf/c1nHx9vMjvzVePCmczWZTVlbWFa3XGHPF9fn7++f53JkzZ9SqVSsVKVJEc+bM0ebNm7V48WJJFy47l6Q2bdro4MGDio6O1pEjR9SiRQvHZfZZWVlq0KCBduzY4fTYu3evunbtKkmaOXOmNm7cqKZNm2rBggWqXr26Nm3adMX7AQCAROgGAOCSAgICVLVqVVWsWDFfM5PXqlVL8fHxOnTokKNt9+7dSk5OVnh4uKMtPj5eR44ccSxv3LhRHh4eql69uoKCglSmTBlt2LDBad2xsbFO68iNj4/PZc8q57fGK1WrVq1ca65evbo8PT3zvZ66detq1apVuT7366+/KikpSePHj9fdd9+tmjVr5nr2vUSJEurdu7fmzJmjKVOm6P3335ck1a9fX/v27VPJkiVVtWpVp8c/f77s9ttv18iRIxUbG6vatWtr7ty5+a4fAIB/InQDAOBC9957r+rWratu3bpp27Zt+vHHH9WzZ09FREQ4XQru5+enXr16aefOnVq/fr2eeeYZdenSRWFhYZKk5557Tq+//roWLFigPXv26Pnnn9eOHTs0ePDgS26/UqVKOn36tFatWqWkpCSlpqYWuMYrNXToUK1atUqvvPKK9u7dq9mzZ+vtt9/OdTK3Sxk9erTmzZun0aNHKy4uTj///LMmTJgg6cJl+T4+Ppo6dap+//13LV26VK+88orT61966SV9/vnn+u2337Rr1y59+eWXjj8mdOvWTaGhoerYsaPWr1+vAwcOaO3atRo8eLAOHz6sAwcOaOTIkdq4caMOHjyoFStWaO/evVf1xwgAwM2N0A0AgAvZbDYtWbJExYoVU/PmzXXvvfeqSpUqWrBggVO/qlWrqnPnzmrbtq1atWql2rVra9q0aY7nn3nmGQ0dOlRDhw5VnTp1tHz5ci1dulTVqlW75PabNm2qAQMG6JFHHlGJEiUcYbUgNV6p+vXr65NPPtH8+fNVu3ZtvfTSSxozZozTzOL5ERkZqYULF2rp0qWqV6+e7rnnHv3www+SLpzBnjVrlhYuXKhatWpp/PjxevPNN51e7+Pjo5EjR6pu3bpq3ry5PD09NX/+fEmS3W7XunXrVKFCBXXu3Fnh4eHq27evzp49q6CgINntdv3666968MEHVb16dT3xxBN6+umn1b9//6t6bwAANy+bKcjNVgAAoMBiYmK0ZMkSx0+MAQCAGxdnugEAAAAAsAihGwAAAAAAi3B5OQAAAAAAFuFMNwAAAAAAFiF0AwAAAABgEUI3AAAAAAAWIXQDAAAAAGARQjcAAAAAABYhdAMAAAAAYBFCNwAAAAAAFiF0AwAAAABgEUI3AAAAAAAW+X/DTOggRfusdwAAAABJRU5ErkJggg==", + "text/plain": [ + "
" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], "source": [ "# Import necessary libraries\n", "import pandas as pd\n", @@ -131,6 +166,7 @@ " return p_wedding_infections, p_wedding_traces\n", "\n", "# Run the simulation 1000 times\n", + "np.random.seed(44)\n", "results = [simulate_event(m) for m in range(1000)]\n", "props_df = pd.DataFrame(results, columns=[\"Infections\", \"Traces\"])\n", "\n", @@ -193,7 +229,7 @@ ], "metadata": { "kernelspec": { - "display_name": "Python 3", + "display_name": "base", "language": "python", "name": "python3" }, @@ -207,7 +243,7 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.13.0" + "version": "3.13.9" } }, "nbformat": 4, diff --git a/02_activities/assignments/a2_survey_design_and_evaluation.md b/02_activities/assignments/a2_survey_design_and_evaluation.md index b4f036f2..95e5e959 100644 --- a/02_activities/assignments/a2_survey_design_and_evaluation.md +++ b/02_activities/assignments/a2_survey_design_and_evaluation.md @@ -40,39 +40,154 @@ For the **Canadian General Social Survey on Giving, Volunteering, and Participat ## Part A - Survey Design: -The number of your chosen topic: `#` +The number of your chosen topic: `1` Describe the purpose of your survey: -``` -write your answer here... -``` +The purpose of this survey is to investigate the underlying reasons for high turnover among entry- and lower-level employees across all departments of the company. +The survey aims to identify key pain points in the employee experience and gather +actionable insights that can inform organizational changes to improve job satisfaction and retention. Describe your target population, sampling frame, sampling units, and observational units: -``` -write your answer here... -``` +Target population: All current entry- and lower-level employees across all +departments, as well as former employees who left the company within the past +12 months. + +Sampling frame: (1) For current employees — the HR database listing all active +entry- and lower-level employees by department. (2) For departed employees — +exit interview records and available contact information for staff who left within +the past 12 months. + +Sampling units: Individual employees (current or recently departed). + +Observational units: Individual employees. + +Sampling strategy: Stratified random sampling, with departments as strata. +This ensures that every department is proportionally represented, which is +important because the turnover problem spans multiple departments and root +causes may differ by team. For departed employees, a census approach (contacting +all available former employees) is used, since this group is small and their +perspective is especially critical for understanding turnover. Your 5-10 question survey: -``` -1. write your question here... -2. write your question here... -3. write your question here... -4. write your question here... -5. write your question here... -6. write your question here... (optional) -7. write your question here... (optional) -8. write your question here... (optional) -9. write your question here... (optional) -10. write your question here... (optional) -``` +1. How long have you worked (or did you work) at this company? + [ ] Less than 6 months [ ] 6–12 months [ ] 1–2 years [ ] More than 2 years + +2. On a scale of 1 to 5, how satisfied are you (or were you) with your overall + work experience at this company? + (1 = Very dissatisfied, 5 = Very satisfied) + +3. What is (or was) the primary reason you are considering leaving (or did leave) + the company? (Select all that apply) + [ ] Compensation [ ] Limited career growth [ ] Poor management + [ ] Work-life balance [ ] Company culture [ ] Better opportunity elsewhere + [ ] Other: ______ + +4. How would you rate your relationship with your direct manager? + (1 = Very poor, 5 = Excellent) + +5. Do you feel (or did you feel) that your compensation and benefits are + fair for your role and responsibilities? + [ ] Yes [ ] No [ ] Unsure + +6. Do you feel (or did you feel) there are clear opportunities for career + advancement within the company? + [ ] Yes [ ] Somewhat [ ] No + +7. How would you rate your work-life balance at this company? + (1 = Very poor, 5 = Excellent) + +8. How would you describe the overall workplace culture? (Open-ended) + +9. What single change would most improve employee satisfaction and retention + at this company? (Open-ended) + +10. Would you recommend this company to a friend as a place to work? + [ ] Yes [ ] No [ ] Maybe ## Part B - Survey Evaluation: Identify and describe survey features: - -``` -write your answer here -``` +1. Sample type: + Stratified random (probability) sample, stratified by province and + Census Metropolitan Area (CMA). A cross-sectional design was used. + +2. Sample size: + Approximately 20,000 respondents completed the survey + (with a larger number initially selected/contacted). + +3. Target population: + All non-institutionalized persons aged 15 years and older living in + private households in Canada's 10 provinces. Excludes residents of + Yukon, Northwest Territories, and Nunavut, and full-time residents + of institutions (e.g., prisons, care homes). + +4. Sampling frame: + Statistics Canada's common telephone frame, which combines landline + and cellular telephone numbers, integrated with the dwelling/Address + Register frame. This replaced older random digit dialing (RDD) methods + and improved coverage. + +5. Survey mode(s): + Two modes were offered: + - CATI (Computer-Assisted Telephone Interviewing) — interviewer-assisted + - rEQ (Respondent-completed Electronic Questionnaire, i.e., online) — + offered for the FIRST TIME in the 2018 cycle + Questionnaire available in both English and French. + +6. Timeline: + Data collection: September to December 2018 + Public Use Microdata File (PUMF) released: January 26, 2021 + +7. Response rate: + Overall response rate: 40.9% + - Regular sample: 42.2% + - Oversample: 37.1% + +8. Weights: + - WGHT_PER: person-level survey weight, adjusted for non-response + and population benchmarks (age, sex, province) + - Bootstrap weights: provided for design-based variance estimation + (accounting for the complex stratified design) + - Record linkage to personal tax files used to supplement income data + (since 2014) + +9. Data processing: + Statistics Canada's standard SSPE (Survey Support Processing Environment) + generalized processing steps were applied. Automated and manual edits + were performed, including consistency, flow, and family edits. The CATI + system included built-in validation and questionnaire flow controls. + +10. Cleaning, imputation, etc.: + - Donor imputation used to address item non-response (e.g., hours + volunteered, donation amounts, income) + - Score functions matched donors and recipients on relevant characteristics + - Income data linked from personal tax records (administrative linkage) + to reduce non-response on income questions + - Approximately 88 variables were subject to imputation in prior cycles; + similar approach applied in 2018 + +11. Sources of error: + - Sampling error: inherent variability due to surveying a sample rather + than the full population; estimated using bootstrap resampling methods + - Non-response error: overall response rate of ~41% means non-response + bias is a significant concern; weights partially correct for this + - Coverage error: exclusion of territories and institutionalized persons + means results are not fully representative of all Canadians + - Measurement error: self-reported data on volunteering hours and + donation amounts may involve recall bias or social desirability bias + +12. Limitations / known biases: + - Excludes Yukon, NWT, Nunavut → results not generalizable to all Canada + - Excludes institutionalized persons (who may have different + volunteering/donating patterns) + - Low response rate (~41%) increases risk of non-response bias + - Introduction of online mode (rEQ) in 2018 affects comparability + with previous cycles (2004, 2007, 2010, 2013) + - Questionnaire was significantly reworked in 2018 (new international + standards, new online tech questions) → results CANNOT be directly + compared to earlier GVP cycles + - Income questions historically have high non-response rates even + after imputation ## Rubric