notebooks/usage-LinearBoost.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "\n",
    "from sklearn.linear_model import *\n",
    "from lineartree import LinearBoostClassifier, LinearBoostRegressor\n",
    "\n",
    "from sklearn.datasets import make_classification, make_regression\n",
    "\n",
    "import warnings\n",
    "warnings.simplefilter('ignore')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# REGRESSION"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((8000, 15), (8000,))"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "n_sample, n_features = 8000, 15\n",
    "X, y = make_regression(n_samples=n_sample, n_features=n_features, n_targets=1, \n",
    "                       n_informative=5, shuffle=True, random_state=33)\n",
    "\n",
    "X.shape, y.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### default configuration"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LinearBoostRegressor(base_estimator=Ridge())"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "regr = LinearBoostRegressor(Ridge(), loss='linear')\n",
    "regr.fit(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((8000, 25), (8000,), 0.9999998985375346)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "regr.transform(X).shape, regr.predict(X).shape, regr.score(X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### square loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LinearBoostRegressor(base_estimator=Ridge(), loss='square', n_estimators=50)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "regr = LinearBoostRegressor(Ridge(), loss='square', n_estimators=50)\n",
    "regr.fit(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((8000, 65), (8000,), 0.9999998980115891)"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "regr.transform(X).shape, regr.predict(X).shape, regr.score(X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### absolute loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LinearBoostRegressor(base_estimator=Ridge(), loss='absolute', n_estimators=50)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "regr = LinearBoostRegressor(Ridge(), loss='absolute', n_estimators=50)\n",
    "regr.fit(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((8000, 65), (8000,), 0.9999998009507592)"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "regr.transform(X).shape, regr.predict(X).shape, regr.score(X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### exponential loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LinearBoostRegressor(base_estimator=Ridge(), loss='exponential',\n",
       "                     n_estimators=50)"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "regr = LinearBoostRegressor(Ridge(), loss='exponential', n_estimators=50)\n",
    "regr.fit(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((8000, 65), (8000,), 0.9999998597027936)"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "regr.transform(X).shape, regr.predict(X).shape, regr.score(X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### multi-target regression with weights "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((8000, 15), (8000, 2))"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "n_sample, n_features = 8000, 15\n",
    "X, y = make_regression(n_samples=n_sample, n_features=n_features, n_targets=2, \n",
    "                       n_informative=5, shuffle=True, random_state=33)\n",
    "W = np.random.uniform(1,3, (n_sample,))\n",
    "\n",
    "X.shape, y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LinearBoostRegressor(base_estimator=Ridge(), n_estimators=50)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "regr = LinearBoostRegressor(Ridge(), loss='linear', n_estimators=50)\n",
    "regr.fit(X, y, W)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((8000, 65), (8000, 2), 0.999999867971615)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "regr.transform(X).shape, regr.predict(X).shape, regr.score(X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# CLASSIFICATION"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((8000, 15), (8000,))"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "n_sample, n_features = 8000, 15\n",
    "X, y = make_classification(n_samples=n_sample, n_features=n_features, n_classes=3, \n",
    "                           n_redundant=4, n_informative=5,\n",
    "                           n_clusters_per_class=1,\n",
    "                           shuffle=True, random_state=33)\n",
    "\n",
    "X.shape, y.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### default configuration "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LinearBoostClassifier(base_estimator=RidgeClassifier())"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf = LinearBoostClassifier(RidgeClassifier(), loss='hamming')\n",
    "clf.fit(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((8000, 25), (8000,), (8000, 3), 0.81775)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf.transform(X).shape, clf.predict(X).shape, clf.predict_proba(X).shape, clf.score(X, y)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### entropy loss "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LinearBoostClassifier(base_estimator=LogisticRegression(), loss='entropy',\n",
       "                      n_estimators=50)"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf = LinearBoostClassifier(LogisticRegression(), loss='entropy', n_estimators=50)\n",
    "clf.fit(X, y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((8000, 65), (8000,), (8000, 3), 0.844)"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "clf.transform(X).shape, clf.predict(X).shape, clf.predict_proba(X).shape, clf.score(X, y)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python [conda env:prova]",
   "language": "python",
   "name": "conda-env-prova-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}