YannAhlgrim/reutlingen-university
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
main
reutlingen-university/1_data_science/6_lernfall/main.ipynb
T
YannAhlgrim bc83f47311 lernfall 6 update
2025-05-23 11:45:30 +02:00

306 lines
9.6 KiB
Plaintext
Raw Permalink Blame History

{
"cells": [
{
"cell_type": "markdown",
"metadata": {},
"source": [
"# KNN Model"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 1. Datensatz laden\n"
]
},
{
"cell_type": "code",
"execution_count": 37,
"metadata": {},
"outputs": [
{
"data": {
"text/html": [
"<div>\n",
"<style scoped>\n",
" .dataframe tbody tr th:only-of-type {\n",
" vertical-align: middle;\n",
" }\n",
"\n",
" .dataframe tbody tr th {\n",
" vertical-align: top;\n",
" }\n",
"\n",
" .dataframe thead th {\n",
" text-align: right;\n",
" }\n",
"</style>\n",
"<table border=\"1\" class=\"dataframe\">\n",
" <thead>\n",
" <tr style=\"text-align: right;\">\n",
" <th></th>\n",
" <th>Vertragsnummer</th>\n",
" <th>Jahr</th>\n",
" <th>Vertragsalter</th>\n",
" <th>Status</th>\n",
" <th>Marktzins</th>\n",
" <th>Tarifzins</th>\n",
" <th>Guthaben</th>\n",
" <th>Sparbeitrag</th>\n",
" <th>Beruf</th>\n",
" <th>Alter</th>\n",
" <th>p</th>\n",
" </tr>\n",
" </thead>\n",
" <tbody>\n",
" <tr>\n",
" <th>0</th>\n",
" <td>10000</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>gekuendigt</td>\n",
" <td>8.7</td>\n",
" <td>8.7</td>\n",
" <td>3933.85</td>\n",
" <td>3619.0</td>\n",
" <td>angestellt</td>\n",
" <td>48</td>\n",
" <td>0.154465</td>\n",
" </tr>\n",
" <tr>\n",
" <th>1</th>\n",
" <td>9999</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>aktiv</td>\n",
" <td>8.7</td>\n",
" <td>8.7</td>\n",
" <td>523.93</td>\n",
" <td>482.0</td>\n",
" <td>angestellt</td>\n",
" <td>28</td>\n",
" <td>0.154465</td>\n",
" </tr>\n",
" <tr>\n",
" <th>2</th>\n",
" <td>9998</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>aktiv</td>\n",
" <td>8.7</td>\n",
" <td>8.7</td>\n",
" <td>7281.81</td>\n",
" <td>6699.0</td>\n",
" <td>selbstaendig</td>\n",
" <td>22</td>\n",
" <td>0.231475</td>\n",
" </tr>\n",
" <tr>\n",
" <th>3</th>\n",
" <td>9997</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>aktiv</td>\n",
" <td>8.7</td>\n",
" <td>8.7</td>\n",
" <td>8905.79</td>\n",
" <td>8193.0</td>\n",
" <td>angestellt</td>\n",
" <td>26</td>\n",
" <td>0.154465</td>\n",
" </tr>\n",
" <tr>\n",
" <th>4</th>\n",
" <td>9996</td>\n",
" <td>1</td>\n",
" <td>1</td>\n",
" <td>aktiv</td>\n",
" <td>8.7</td>\n",
" <td>8.7</td>\n",
" <td>8566.65</td>\n",
" <td>7881.0</td>\n",
" <td>angestellt</td>\n",
" <td>30</td>\n",
" <td>0.154465</td>\n",
" </tr>\n",
" </tbody>\n",
"</table>\n",
"</div>"
],
"text/plain": [
" Vertragsnummer Jahr Vertragsalter Status Marktzins Tarifzins \\\n",
"0 10000 1 1 gekuendigt 8.7 8.7 \n",
"1 9999 1 1 aktiv 8.7 8.7 \n",
"2 9998 1 1 aktiv 8.7 8.7 \n",
"3 9997 1 1 aktiv 8.7 8.7 \n",
"4 9996 1 1 aktiv 8.7 8.7 \n",
"\n",
" Guthaben Sparbeitrag Beruf Alter p \n",
"0 3933.85 3619.0 angestellt 48 0.154465 \n",
"1 523.93 482.0 angestellt 28 0.154465 \n",
"2 7281.81 6699.0 selbstaendig 22 0.231475 \n",
"3 8905.79 8193.0 angestellt 26 0.154465 \n",
"4 8566.65 7881.0 angestellt 30 0.154465 "
]
},
"execution_count": 37,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"import pandas as pd\n",
"\n",
"data = pd.read_csv('../3_lernfall/VertragsdatenABCAG.csv', sep=',', decimal='.')\n",
"data.head()"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 2. Datenvorverarbeitung"
]
},
{
"cell_type": "code",
"execution_count": 38,
"metadata": {},
"outputs": [
{
"data": {
"text/plain": [
"array([[1.5611106971092041, -0.9658806326457345, 0.7729934357656066,\n",
" 0.19511754814402793, -0.47513581897659896, 1.389983329807085,\n",
" 0.5164540311483795, 0, 0],\n",
" [0.7340234916233305, 1.7968467730396942, 0.42492374363524343,\n",
" 1.422200684147216, 3.3738011771456753, 1.3529959365231317,\n",
" -0.16122757685231498, 0, 0],\n",
" [-1.74723812483429, -0.5712052889763876, 0.8165021472819023,\n",
" 1.422200684147216, -0.3993293859693223, -0.0075120485158652085,\n",
" 0.008192825147858662, 0, 0],\n",
" [0.3204798888803937, 0.6128207420316534, -0.09718079456030153,\n",
" -0.5988774222109761, 0.998529481910825, 1.1900420434702434,\n",
" -0.7541989838529227, 0, 1],\n",
" [-0.5066073166054799, 0.2181453983623064, -1.315424717016573,\n",
" 0.23120822861471013, -0.037511869651474364, -0.19977670605792403,\n",
" 1.0247152371489006, 0, 0]], dtype=object)"
]
},
"execution_count": 38,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"from sklearn.preprocessing import StandardScaler\n",
"from sklearn.model_selection import train_test_split\n",
"\n",
"# Stichprobe der Daten\n",
"data = data.sample(frac=0.3, random_state=11)\n",
"\n",
"# One hot encoding für die Spalten \"Beruf\" und \"Status\"\n",
"data = pd.get_dummies(data, columns=['Beruf', 'Status'], drop_first=True)\n",
"\n",
"# Zielvariable und Merkmale definieren\n",
"y = data['Status_gekuendigt']\n",
"X = data.drop(['Status_gekuendigt', 'Status_aktiv', 'Status_verstorben', 'Vertragsnummer', 'p'], axis=1)\n",
"\n",
"# Nur numerische Spalten skalieren, Dummy-Variablen nicht skalieren\n",
"num_cols = X.select_dtypes(include=['int64', 'float64']).columns\n",
"dummy_cols = X.select_dtypes(include=['bool']).columns\n",
"\n",
"scaler = StandardScaler()\n",
"X_num_scaled = scaler.fit_transform(X[num_cols])\n",
"X_scaled = X.copy()\n",
"X_scaled[num_cols] = X_num_scaled\n",
"X_scaled = X_scaled.values\n",
"\n",
"# Dummy-Variablen in 0 und 1 umwandeln\n",
"for col in dummy_cols:\n",
" X_scaled[:, X.columns.get_loc(col)] = X[col].astype(int)\n",
"\n",
"# Train-Test-Split\n",
"X_train, X_test, y_train, y_test = train_test_split(X_scaled, y, test_size=0.2, random_state=22)\n",
"\n",
"# Ersten 5 Zeilen der Trainingsdaten anzeigen\n",
"X_train[:5]"
]
},
{
"cell_type": "markdown",
"metadata": {},
"source": [
"## 3. KNN Modell implementieren und MSE berechnen"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
"from sklearn.neighbors import KNeighborsClassifier\n",
"from sklearn.metrics import mean_squared_error\n",
"import matplotlib.pyplot as plt\n",
"\n",
"# Werte für k definieren\n",
"k_values = list(range(1, 101, 1))\n",
"train_mse = []\n",
"test_mse = []\n",
"\n",
"for k in k_values:\n",
" model = KNeighborsClassifier(n_neighbors=k)\n",
" model.fit(X_train, y_train)\n",
" y_train_pred = model.predict_proba(X_train)\n",
" y_test_pred = model.predict_proba(X_test)\n",
" train_mse.append(mean_squared_error(y_train, y_train_pred[:, 1]))\n",
" test_mse.append(mean_squared_error(y_test, y_test_pred[:, 1]))\n",
"\n",
"# mean var (p*(1-p)).mean()\n",
"# oder: np.var(y - p)\n",
"\n",
"# Ergebnisse visualisieren\n",
"plt.figure(figsize=(10, 6))\n",
"plt.plot(k_values, train_mse, label='Train MSE', marker='o')\n",
"plt.plot(k_values, test_mse, label='Test MSE', marker='o')\n",
"\n",
"# vertikale Linie für den besten k-Wert einzeichnen\n",
"min_test_mse = min(test_mse)\n",
"best_k = k_values[test_mse.index(min_test_mse)]\n",
"plt.axvline(x=best_k, color='red', linestyle='--', label=f'Best k = {best_k}')\n",
"\n",
"plt.title('MSE für verschiedene Werte von k')\n",
"plt.xlabel('Wert von k')\n",
"plt.ylabel('Mean Squared Error')\n",
"plt.xscale('log')\n",
"plt.legend()\n",
"plt.grid()\n",
"plt.show()"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "hft_ml",
"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": "3.11.10"
}
},
"nbformat": 4,
"nbformat_minor": 4
}
Reference in New Issue View Git Blame Copy Permalink