{ "cells": [ { "cell_type": "markdown", "id": "9d2b1489-2c3c-4097-9733-78c5cfdd72ab", "metadata": {}, "source": [ "### 과대적합과 과소적합\n", "\n", "\n", "#### 과대적합(overfitting)\n", "* 훈련데이터에만 잘 맞고 테스트데이터에는 낮은 정답률이 나오는 경우\n", "\n", "\n", "\n", "#### 과소적합(underfitting)\n", "* 훈련이 제대로 이루어지지 않아 테스트 데이터보다 훈련데이터의 정답률이 낮은 경우\n", "* 훈련, 테스트데이터 모두 정답률이 낮은 경우\n", "* 훈련데이터의 개수가 적을 경우\n", "\n", "\n", "\n", "---ht=\"\" />\r\n", "\r\n", "---" ] }, { "cell_type": "markdown", "id": "c24afdc0-02ee-4e95-accf-7899477c5155", "metadata": {}, "source": [ "### 스케일링\n", "* 수치형 데이터들의 값의 범위가 클 경우 사용한다.(연산 속도 증가)\n", " - 예를 들어 주식의 가격을 생각했을 경우 100원의 1%와 100000원의 1%를 계산하고자 한다면 같은 1%연산이지만 숫자가 클 경우 연산속도에 영향을 미치게 된다.\n", "* 서로 다른 변수의 값 범위를 일정한 수준으로 맞추고자 하는 경우\n", "* 대표적 스케일링 클래스는 StandardScaler, MinMaxScaler이 있다\n", "* 사용 이유\n", " - 연산속도를 올릴 수 있다\n", " - 확률을 높일 수 있다\n", " - 과대, 과소적합을 조금은 해결할 수 있다\n", "* 참고\n", " - 이상치가 있을 경우 범위의 차이가 커지게 되므로 이상치는 제거해야 한다\n", " - RandomForest는 따로 스케일링 할 필요가 없다.(tree계열은 따로 스케일 할 필요가 없다)" ] }, { "cell_type": "markdown", "id": "087379fa-55cd-422e-8c73-a45646316f3f", "metadata": {}, "source": [ "#### StandardScaler(label값은 스케일 하지 않는다)\n", "* 사이킷런에서 서포트벡터머신, 선형회귀, 로지스틱 회귀에 적용하는 것이 예측 성능 향상에 중요한 요소가 된다.\n", "* 각 특성의 평균을 0, 편차를 1로 변경하여 정규분포로 만드는 것이다\n", " - 편차(deviation)는 예측값과 평균의 차이\n", " \n", "\n", "\n", "#### MinMaxScaler(label값은 스케일 하지 않는다)\n", "* 모든 특성이 정확하게 0과 1사이에 위치하도록 데이터를 변경\n", "* k-means, k-nearest neighbor등의 알고리즘에 좋은 성능을 보인다은 성능을 보인다" ] }, { "cell_type": "code", "execution_count": 3, "id": "5445eb41-4eb0-4c7c-841c-59ba4ba15c04", "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", "
daumnaver
021
142
263
384
4105
\n", "
" ], "text/plain": [ " daum naver\n", "0 2 1\n", "1 4 2\n", "2 6 3\n", "3 8 4\n", "4 10 5" ] }, "execution_count": 3, "metadata": {}, "output_type": "execute_result" } ], "source": [ "import pandas as pd\n", "from sklearn.preprocessing import StandardScaler\n", "\n", "# 가상의 영화 평점 데이터\n", "movie = {'daum':[2,4,6,8,10], 'naver':[1,2,3,4,5]}\n", "mv = pd.DataFrame(movie)\n", "mv" ] }, { "cell_type": "code", "execution_count": 13, "id": "57574836-5b5e-433e-8dd2-fc858a6b6c01", "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", "
daumnaver
0-1.414214e+00-1.414214e+00
1-7.071068e-01-7.071068e-01
24.440892e-174.440892e-17
37.071068e-017.071068e-01
41.414214e+001.414214e+00
\n", "
" ], "text/plain": [ " daum naver\n", "0 -1.414214e+00 -1.414214e+00\n", "1 -7.071068e-01 -7.071068e-01\n", "2 4.440892e-17 4.440892e-17\n", "3 7.071068e-01 7.071068e-01\n", "4 1.414214e+00 1.414214e+00" ] }, "execution_count": 13, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# standardscaler 를 사용해서\n", "# 두 데이터 간의 범위를 줄인다\n", "st = StandardScaler()\n", "scaled = st.fit_transform(mv)\n", "mv = pd.DataFrame(data = scaled, columns=['daum', 'naver'])\n", "mv" ] }, { "cell_type": "code", "execution_count": 12, "id": "7e844027-5c1b-451e-8c7e-fc2aaeea9a94", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "1.0" ] }, "execution_count": 12, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# round() : 소숫점 범위를 잘라준다\n", "round(scaled.std(), 2)" ] }, { "cell_type": "code", "execution_count": null, "id": "7198e77d-c788-443c-860c-3e928ef91b48", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "547f6b5c-4179-40f9-ba0a-de1612dd1df5", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": 16, "id": "95d3adff-a0f4-4526-9b36-ac5d0bd9d67a", "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", "
daumnaver
021
142
263
384
4105
\n", "
" ], "text/plain": [ " daum naver\n", "0 2 1\n", "1 4 2\n", "2 6 3\n", "3 8 4\n", "4 10 5" ] }, "execution_count": 16, "metadata": {}, "output_type": "execute_result" } ], "source": [ "from sklearn.preprocessing import MinMaxScaler\n", "mv = pd.DataFrame(movie)\n", "mv" ] }, { "cell_type": "code", "execution_count": 18, "id": "964a8dee-0318-4c5a-8c65-f93a772fd7cc", "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", "
db
00.000.00
10.250.25
20.500.50
30.750.75
41.001.00
\n", "
" ], "text/plain": [ " d b\n", "0 0.00 0.00\n", "1 0.25 0.25\n", "2 0.50 0.50\n", "3 0.75 0.75\n", "4 1.00 1.00" ] }, "execution_count": 18, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# MinMaxScaler 를 사용해\n", "# 다른 두 값을 0 ~ 1 사이의 값으로\n", "# 바꿔준다\n", "m = MinMaxScaler()\n", "min_max_mv = m.fit_transform(mv)\n", "pd.DataFrame(min_max_mv, columns=['d', 'b'])" ] }, { "cell_type": "code", "execution_count": null, "id": "00701a04-5d74-4175-99fe-bce9a7607f81", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "55afd9d5-f413-452d-8763-85d617a701cf", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": 19, "id": "fb8c4fb0-dfb4-471b-9cc5-c3844de41c1a", "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", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", " \n", "
agedurationcampaignpdayspreviousy
0582611-100
1441511-100
233761-100
347921-100
4331981-100
\n", "
" ], "text/plain": [ " age duration campaign pdays previous y\n", "0 58 261 1 -1 0 0\n", "1 44 151 1 -1 0 0\n", "2 33 76 1 -1 0 0\n", "3 47 92 1 -1 0 0\n", "4 33 198 1 -1 0 0" ] }, "execution_count": 19, "metadata": {}, "output_type": "execute_result" } ], "source": [ "df = pd.read_csv(\"../data_set/5.스케일링/bank_train_clean.csv\")\n", "df.head()" ] }, { "cell_type": "code", "execution_count": 20, "id": "b9d44017-9f58-4d58-a7fa-6587dda3a5d7", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "Index(['age', 'duration', 'campaign', 'pdays', 'previous', 'y'], dtype='object')" ] }, "execution_count": 20, "metadata": {}, "output_type": "execute_result" } ], "source": [ "df.columns" ] }, { "cell_type": "code", "execution_count": 21, "id": "d33594bd-e4c6-4f35-b070-1eecc10c854e", "metadata": {}, "outputs": [], "source": [ "features = ['age', 'duration', 'campaign', 'pdays', 'previous']\n", "label = 'y'\n", "\n", "X, y = df[features], df[label]" ] }, { "cell_type": "code", "execution_count": 22, "id": "0128f64b-ace1-4387-9073-310acd63827b", "metadata": {}, "outputs": [], "source": [ "from sklearn.model_selection import train_test_split\n", "\n", "# 학습용 70%, 테스트용 30%\n", "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3)" ] }, { "cell_type": "code", "execution_count": 26, "id": "52465bbc-e3af-4594-8ec8-9d090a8aa5c2", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "학습 데이터 : 0.9923847442095617\n", "test 데이터 : 0.8815246240047184\n" ] } ], "source": [ "from sklearn.ensemble import RandomForestClassifier\n", "\n", "# 머신 생성\n", "rfc = RandomForestClassifier()\n", "\n", "# 머신 학습\n", "rfc.fit(X_train, y_train);\n", "\n", "print( \"학습 데이터 : \", rfc.score(X_train, y_train) )\n", "\n", "# 학습된 머신 테스트 결과\n", "print( \"test 데이터 : \", rfc.score(X_test, y_test) )\n", "\n", "# 학습 데이터의 결과와 테스트 결과의 차이가 커지면\n", "# 과적합 되었다라고 말한다" ] }, { "cell_type": "code", "execution_count": null, "id": "f32f76e0-dce7-4bfc-85c0-06039fe6dc02", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "7d323808-1d19-4198-bee1-398bda3b5547", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": 28, "id": "15b62827-c8a7-4a45-b909-17b056bafad1", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "학습 데이터 : 0.9139570891395709\n", "test 데이터 : 0.8846210557357712\n" ] } ], "source": [ "from sklearn.neighbors import KNeighborsClassifier\n", "\n", "# 머신 생성\n", "kn = KNeighborsClassifier()\n", "\n", "# 머신 학습\n", "kn.fit(X_train, y_train)\n", "\n", "print( \"학습 데이터 : \", kn.score(X_train, y_train) )\n", "\n", "# 학습된 머신 테스트 결과\n", "print( \"test 데이터 : \", kn.score(X_test, y_test) )\n", "\n", "# 학습 데이터와 test 데이터 테스트 결과의\n", "# 차이 폭이 적으므로 좋은 알고리즘이다...." ] }, { "cell_type": "code", "execution_count": null, "id": "c744cdff-c398-492a-9ff2-2cb4ed9e9c87", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "3ee38aa7-e051-49a4-b18c-46dbf5638f1f", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": 30, "id": "49a313b4-1ef8-414d-bf2a-f8de8cfabfe3", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([[ 1.60696496, 0.0110161 , -0.56935064, -0.41145311, -0.25194037],\n", " [ 0.28852927, -0.41612696, -0.56935064, -0.41145311, -0.25194037],\n", " [-0.74738448, -0.70736086, -0.56935064, -0.41145311, -0.25194037],\n", " ...,\n", " [ 2.92540065, 3.37379688, 0.72181052, 1.43618859, 1.05047333],\n", " [ 1.51279098, 0.97014641, 0.39902023, -0.41145311, -0.25194037],\n", " [-0.37068857, 0.39932797, -0.24656035, 1.4761376 , 4.52357654]])" ] }, "execution_count": 30, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# 데이터 스케일링\n", "sc = StandardScaler()\n", "sc.fit( X )\n", "X2 = sc.transform( X )\n", "X2" ] }, { "cell_type": "code", "execution_count": 31, "id": "fb66e42d-30ca-4681-9dbf-6c5660cedc9f", "metadata": {}, "outputs": [], "source": [ "X_train, X_test, y_train, y_test = train_test_split(X2, y, test_size=0.3)" ] }, { "cell_type": "code", "execution_count": 35, "id": "ed97e375-5588-4710-9423-0020b6c9977b", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "학습 데이터 : 0.9919423642051379\n", "test 데이터 : 0.8757003833677381\n" ] } ], "source": [ "# 머신 생성\n", "rfc = RandomForestClassifier()\n", "\n", "# 머신 학습\n", "rfc.fit(X_train, y_train);\n", "\n", "print( \"학습 데이터 : \", rfc.score(X_train, y_train) )\n", "\n", "# 학습된 머신 테스트 결과\n", "print( \"test 데이터 : \", rfc.score(X_test, y_test) )\n", "\n", "# 스케일링된 데이터로 다시 테스트" ] }, { "cell_type": "code", "execution_count": 36, "id": "99b350fe-2aa6-437c-88ad-6a065baee8ec", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "학습 데이터 : 0.9128195405567668\n", "test 데이터 : 0.8824830433500442\n" ] } ], "source": [ "# 머신 생성\n", "kn = KNeighborsClassifier()\n", "\n", "# 머신 학습\n", "kn.fit(X_train, y_train)\n", "\n", "print( \"학습 데이터 : \", kn.score(X_train, y_train) )\n", "\n", "# 학습된 머신 테스트 결과\n", "print( \"test 데이터 : \", kn.score(X_test, y_test) )\n", "\n", "# 스케일링된 데이터로 다시 테스트" ] }, { "cell_type": "markdown", "id": "00f394cc-9693-4089-9299-0a950f918ff6", "metadata": {}, "source": [ "### 스케일링을 사용하면 과적합을 줄여줄 수도 있다" ] }, { "cell_type": "code", "execution_count": null, "id": "ce16806f-c1d6-42c7-b7ec-5457c8b8619f", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "bcc20339-f56d-4d1f-a803-b1772eac2e96", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": 41, "id": "d337b3b2-db6a-4357-8f66-c4030eab0cb6", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "array([[0.51948052, 0.05307035, 0. , 0. , 0. ],\n", " [0.33766234, 0.03070354, 0. , 0. , 0. ],\n", " [0.19480519, 0.01545344, 0. , 0. , 0. ],\n", " ...,\n", " [0.7012987 , 0.22915819, 0.06451613, 0.21215596, 0.01090909],\n", " [0.50649351, 0.10329402, 0.0483871 , 0. , 0. ],\n", " [0.24675325, 0.07340382, 0.01612903, 0.21674312, 0.04 ]])" ] }, "execution_count": 41, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# fit_transform() 을 사용하여 스케일링\n", "X2 = m.fit_transform( X )\n", "X2" ] }, { "cell_type": "code", "execution_count": 38, "id": "0bee5ddb-a89b-4cec-840b-37f3b8c9f262", "metadata": {}, "outputs": [], "source": [ "X_train, X_test, y_train, y_test = train_test_split(X2, y, test_size=0.3)" ] }, { "cell_type": "code", "execution_count": 39, "id": "404054c2-6595-4795-a1b9-7ddefb8fd678", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "학습 데이터 : 0.9916895756311814\n", "test 데이터 : 0.8786493659687408\n" ] } ], "source": [ "# 머신 생성\n", "rfc = RandomForestClassifier()\n", "\n", "# 머신 학습\n", "rfc.fit(X_train, y_train);\n", "\n", "print( \"학습 데이터 : \", rfc.score(X_train, y_train) )\n", "\n", "# 학습된 머신 테스트 결과\n", "print( \"test 데이터 : \", rfc.score(X_test, y_test) )\n", "\n", "# 스케일링된 데이터로 다시 테스트" ] }, { "cell_type": "code", "execution_count": 40, "id": "fce36627-acea-4f34-9d22-3527fe4d65b6", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "학습 데이터 : 0.9158846020159889\n", "test 데이터 : 0.8817457976997936\n" ] } ], "source": [ "# 머신 생성\n", "kn = KNeighborsClassifier()\n", "\n", "# 머신 학습\n", "kn.fit(X_train, y_train)\n", "\n", "print( \"학습 데이터 : \", kn.score(X_train, y_train) )\n", "\n", "# 학습된 머신 테스트 결과\n", "print( \"test 데이터 : \", kn.score(X_test, y_test) )\n", "\n", "# 스케일링된 데이터로 다시 테스트" ] }, { "cell_type": "code", "execution_count": null, "id": "334c6973-096f-4b0f-b1c3-42b7c3f75b40", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "294ae389-831a-4806-8adb-4296d30954b9", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": 43, "id": "89824346-3e40-4093-8975-bee674e90fae", "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", " \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", "
agedurationcampaignpdayspreviousy
age1.000000-0.0046480.004760-0.0237580.0012880.025155
duration-0.0046481.000000-0.084570-0.0015650.0012030.394521
campaign0.004760-0.0845701.000000-0.088628-0.032855-0.073172
pdays-0.023758-0.001565-0.0886281.0000000.4548200.103621
previous0.0012880.001203-0.0328550.4548201.0000000.093236
y0.0251550.394521-0.0731720.1036210.0932361.000000
\n", "
" ], "text/plain": [ " age duration campaign pdays previous y\n", "age 1.000000 -0.004648 0.004760 -0.023758 0.001288 0.025155\n", "duration -0.004648 1.000000 -0.084570 -0.001565 0.001203 0.394521\n", "campaign 0.004760 -0.084570 1.000000 -0.088628 -0.032855 -0.073172\n", "pdays -0.023758 -0.001565 -0.088628 1.000000 0.454820 0.103621\n", "previous 0.001288 0.001203 -0.032855 0.454820 1.000000 0.093236\n", "y 0.025155 0.394521 -0.073172 0.103621 0.093236 1.000000" ] }, "execution_count": 43, "metadata": {}, "output_type": "execute_result" } ], "source": [ "# corr() : 데이터 간의 상관관계를 확인\n", "df.corr()" ] }, { "cell_type": "code", "execution_count": 45, "id": "58c68179-3c24-4736-b0c1-165cfccaf893", "metadata": {}, "outputs": [ { "data": { "text/plain": [ "" ] }, "execution_count": 45, "metadata": {}, "output_type": "execute_result" }, { "data": { "image/png": "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", "text/plain": [ "
" ] }, "metadata": {}, "output_type": "display_data" } ], "source": [ "import seaborn as sns\n", "\n", "# annot = True : 상관관계 숫자를 출력\n", "sns.heatmap( df.corr(), annot=True )" ] }, { "cell_type": "code", "execution_count": 55, "id": "485f61ba-0a09-45ad-b515-63f26e8b1e72", "metadata": {}, "outputs": [], "source": [ "# features = ['age', 'duration', 'campaign', 'pdays', 'previous']\n", "# features = ['age', 'duration', 'campaign', 'previous']\n", "# features = ['age', 'campaign', 'previous']\n", "features = ['age', 'campaign', 'previous']\n", "label = 'y'\n", "X, y = df[features], df[label]\n", "\n", "sc = StandardScaler()\n", "sc.fit( X )\n", "X_scaler = sc.transform( X )\n", "\n", "X_train, X_test, y_train, y_test = train_test_split(X_scaler, y, test_size=0.3)" ] }, { "cell_type": "code", "execution_count": 56, "id": "eebfdd78-68b5-42d7-95b9-8d27335b239e", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "학습 데이터 : 0.8956615160994723\n", "test 데이터 : 0.8768799764081392\n", "학습 데이터 : 0.8799886245141719\n", "test 데이터 : 0.8695812444706577\n" ] } ], "source": [ "rfc = RandomForestClassifier()\n", "rfc.fit(X_train, y_train);\n", "print( \"학습 데이터 : \", rfc.score(X_train, y_train) )\n", "print( \"test 데이터 : \", rfc.score(X_test, y_test) )\n", "\n", "kn = KNeighborsClassifier()\n", "kn.fit(X_train, y_train)\n", "print( \"학습 데이터 : \", kn.score(X_train, y_train) )\n", "print( \"test 데이터 : \", kn.score(X_test, y_test) )" ] } ], "metadata": { "kernelspec": { "display_name": "Python 3 (ipykernel)", "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": "3.11.7" } }, "nbformat": 4, "nbformat_minor": 5 }