{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "# Causal Structures - Prediction" ] }, { "cell_type": "code", "execution_count": 1, "metadata": {}, "outputs": [], "source": [ "%%capture\n", "# execute the creation & training notebook first\n", "%run \"02-01-creation_and_training.ipynb\"" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "In the [previous section](./02-01-creation_and_training.ipynb) we created and trained our causal structure.\n", "We can now make predictions. Let's create some test data first." ] }, { "cell_type": "code", "execution_count": 2, "metadata": {}, "outputs": [], "source": [ "test_data_a = {\"(a)\": np.linspace(4.5, 5.5, 100)}\n", "prediction = causal_structure.predict(data=test_data_a)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Since we passed a dictionary as data to the predict method, the result is also a dictionary." ] }, { "cell_type": "code", "execution_count": 3, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "pandas.core.frame.DataFrame" ] }, "metadata": {}, "output_type": "display_data" }, { "data": { "text/plain": [ "Index(['(a)', '(b|a)', '(c|a,b)'], dtype='object')" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "display(type(prediction))\n", "display(prediction.keys())" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "If we pass a pandas data frame we get a pandas data frame in return." ] }, { "cell_type": "code", "execution_count": 4, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
(a)(b|a)(c|a,b)
04.500000-3.95014245.574075
14.510101-4.43672045.443720
24.520202-4.92757445.272895
34.530303-5.40899745.130555
44.540404-5.90711744.996075
\n", "
" ], "text/plain": [ " (a) (b|a) (c|a,b)\n", "0 4.500000 -3.950142 45.574075\n", "1 4.510101 -4.436720 45.443720\n", "2 4.520202 -4.927574 45.272895\n", "3 4.530303 -5.408997 45.130555\n", "4 4.540404 -5.907117 44.996075" ] }, "execution_count": 4, "metadata": {}, "output_type": "execute_result" } ], "source": [ "test_data_a = pd.DataFrame(data=test_data_a)\n", "prediction = causal_structure.predict(data=test_data_a)\n", "prediction.head()" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Note that even though the column '(c|a,b)' depends on both '(a)' and '(b|a)' and we only provided data for '(a)' we get a prediction for '(c|a,b)'. This prediction is based on averaging the possible values of '(b|a)' given the provided data for '(a)'." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We can plot the prediction and compare it to the training data." ] }, { "cell_type": "code", "execution_count": 5, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Text(0, 0.5, '(c|a,b)')" ] }, "execution_count": 5, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "pl.figure(figsize=(10,3.5))\n", "fig = pl.subplot(1,2,1)\n", "fig.plot(prediction[\"(a)\"], prediction[\"(b|a)\"], color=\"red\")\n", "fig.scatter(data[\"(a)\"], data[\"(b|a)\"], marker=\"x\", color=\"k\")\n", "fig.set_xlabel(\"(a)\")\n", "fig.set_ylabel(\"(b|a)\")\n", "fig = pl.subplot(1,2,2)\n", "fig.plot(prediction[\"(a)\"], prediction[\"(c|a,b)\"], color=\"red\")\n", "fig.scatter(data[\"(a)\"], data[\"(c|a,b)\"], marker=\"x\", color=\"k\")\n", "fig.set_xlabel(\"(a)\")\n", "fig.set_ylabel(\"(c|a,b)\")" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Prediction uncertainties" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "If we want to see the uncertainty margin of the prediction we have to ask the ``predict`` method for the standard deviation." ] }, { "cell_type": "code", "execution_count": 6, "metadata": {}, "outputs": [], "source": [ "prediction_mean, prediction_std = causal_structure.predict(\n", " data=test_data_a, return_std=True)" ] }, { "cell_type": "code", "execution_count": 7, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Text(0, 0.5, '(c|a,b)')" ] }, "execution_count": 7, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "pl.figure(figsize=(10,3.5))\n", "fig = pl.subplot(1,2,1)\n", "fig.plot(prediction_mean[\"(a)\"], prediction_mean[\"(b|a)\"], color=\"red\")\n", "fig.fill_between(prediction_mean[\"(a)\"],\n", " (prediction_mean - prediction_std)[\"(b|a)\"],\n", " (prediction_mean + prediction_std)[\"(b|a)\"],\n", " color=\"red\", alpha=0.5)\n", "fig.scatter(data[\"(a)\"], data[\"(b|a)\"], marker=\"x\", color=\"k\")\n", "fig.set_xlabel(\"(a)\")\n", "fig.set_ylabel(\"(a|b)\")\n", "fig = pl.subplot(1,2,2)\n", "fig.plot(prediction_mean[\"(a)\"], prediction_mean[\"(c|a,b)\"], color=\"red\")\n", "fig.fill_between(prediction_mean[\"(a)\"],\n", " (prediction_mean - prediction_std)[\"(c|a,b)\"],\n", " (prediction_mean + prediction_std)[\"(c|a,b)\"],\n", " color=\"red\", alpha=0.5)\n", "fig.scatter(data[\"(a)\"], data[\"(c|a,b)\"], marker=\"x\", color=\"k\")\n", "fig.set_xlabel(\"(a)\")\n", "fig.set_ylabel(\"(c|a,b)\")" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The uncertainty margin consists of two contributions. The learned variance of the quadratic regressions and the uncertainty of the regression parameters.\n", "\n", "With the regression equation,\n", "\n", "$y(x) = a \\cdot x + b \\cdot x^2 + c + \\xi$,\n", "\n", "the learned variance describes the strength of the random noise $\\xi$ and the uncertainty of the regression parameters quantifies that with finite data we know $a$, $b$, and $c$ only with a finite precision." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "### Backwards prediction" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "We can also evaluate the inverse prediction, e.g. predicting '(b|a)' and '(a)' from '(c|a,b)'.\n", "We just have to provide data for '(c|a,b)' to the predict method and the causal structure will solve the inverse model." ] }, { "cell_type": "code", "execution_count": 8, "metadata": {}, "outputs": [ { "data": { "text/html": [ "
\n", "\n", "\n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
(a)(b|a)(c|a,b)
05.199343-35.29070636.000000
15.195043-35.02251136.080808
25.190741-34.83274336.161616
35.186438-34.65321636.242424
45.182134-34.41384536.323232
\n", "
" ], "text/plain": [ " (a) (b|a) (c|a,b)\n", "0 5.199343 -35.290706 36.000000\n", "1 5.195043 -35.022511 36.080808\n", "2 5.190741 -34.832743 36.161616\n", "3 5.186438 -34.653216 36.242424\n", "4 5.182134 -34.413845 36.323232" ] }, "execution_count": 8, "metadata": {}, "output_type": "execute_result" } ], "source": [ "test_data_c = pd.DataFrame(data={'(c|a,b)': np.linspace(36, 44, 100)})\n", "\n", "prediction_mean, prediction_std = causal_structure.predict(\n", " data=test_data_c, return_std=True)\n", "\n", "prediction_mean.head()" ] }, { "cell_type": "code", "execution_count": 9, "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Text(0, 0.5, '(b|a)')" ] }, "execution_count": 9, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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\n", "text/plain": [ "
" ] }, "metadata": { "needs_background": "light" }, "output_type": "display_data" } ], "source": [ "pl.figure(figsize=(10,3.5))\n", "fig = pl.subplot(1,2,1)\n", "fig.plot(prediction_mean[\"(c|a,b)\"], prediction_mean[\"(a)\"], color=\"red\")\n", "fig.fill_between(prediction_mean[\"(c|a,b)\"],\n", " (prediction_mean - prediction_std)[\"(a)\"],\n", " (prediction_mean + prediction_std)[\"(a)\"],\n", " color=\"red\", alpha=0.5)\n", "fig.scatter(data[\"(c|a,b)\"], data[\"(a)\"], marker=\"x\", color=\"k\")\n", "fig.set_xlabel(\"(c|a,b)\")\n", "fig.set_ylabel(\"(a)\")\n", "fig = pl.subplot(1,2,2)\n", "fig.plot(prediction_mean[\"(c|a,b)\"], prediction_mean[\"(b|a)\"], color=\"red\")\n", "fig.fill_between(prediction_mean[\"(c|a,b)\"],\n", " (prediction_mean - prediction_std)[\"(b|a)\"],\n", " (prediction_mean + prediction_std)[\"(b|a)\"],\n", " color=\"red\", alpha=0.5)\n", "fig.scatter(data[\"(c|a,b)\"], data[\"(b|a)\"], marker=\"x\", color=\"k\")\n", "fig.set_xlabel(\"(c|a,b)\")\n", "fig.set_ylabel(\"(b|a)\")" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "The plots of the backwards prediction show us that the backwards prediction usually is a little bit conservative. The predictions stay closer to the average value than the data points.\n", "\n", "This is common behavior in inverse modelling and has to do with the fact that a high value of '(c|a,b)' can have multiple causes, a high value in '(a)', a high value in '(a|b)' or a high value of the random noise contribution. This effect is explained in more detail in the [inheritance example](./../01_introduction/02_inheritance_example.ipynb) in the core-documentation." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "Apart from predicting values a causal structure can evaluate objectives.\n", "We will see what they are in the [next section](./02-03-objectives_intro.ipynb)." ] } ], "metadata": { "kernelspec": { "display_name": "Python 3", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.8.8" } }, "nbformat": 4, "nbformat_minor": 4 }