Preprocessing data

Supervised Learning with scikit-learn

George Boorman

Core Curriculum Manager, DataCamp

scikit-learn requirements

  • Numeric data
  • No missing values

 

  • With real-world data:
    • This is rarely the case
    • We will often need to preprocess our data first
Supervised Learning with scikit-learn

Dealing with categorical features

  • scikit-learn will not accept categorical features by default

  • Need to convert categorical features into numeric values

  • Convert to binary features called dummy variables

    • 0: Observation was NOT that category

    • 1: Observation was that category

Supervised Learning with scikit-learn

Dummy variables

list of eleven genres including Electronic, Hip-Hop, and Rock

Supervised Learning with scikit-learn

Dummy variables

Grid with genre values and column names and zeroes and ones as values, where 1 means the song is that genre, and 0 means it isn’t

Supervised Learning with scikit-learn

Dummy variables

Grid of ones and zeroes without a column for Rock genre

Supervised Learning with scikit-learn

Dealing with categorical features in Python

  • scikit-learn: OneHotEncoder()

  • pandas: get_dummies()

Supervised Learning with scikit-learn

Music dataset

  • popularity: Target variable
  • genre: Categorical feature
print(music.info())

     popularity    acousticness    danceability    ...    tempo         valence    genre
0    41.0          0.6440          0.823           ...    102.619000    0.649      Jazz
1    62.0          0.0855          0.686           ...    173.915000    0.636      Rap
2    42.0          0.2390          0.669           ...    145.061000    0.494      Electronic
3    64.0          0.0125          0.522           ...    120.406497    0.595      Rock
4    60.0          0.1210          0.780           ...    96.056000     0.312      Rap
Supervised Learning with scikit-learn

EDA w/ categorical feature

Box plot showing popularity for each genre

Supervised Learning with scikit-learn

Encoding dummy variables

import pandas as pd

music_df = pd.read_csv('music.csv')

music_dummies = pd.get_dummies(music_df["genre"], drop_first=True)

print(music_dummies.head())

     Anime    Blues    Classical    Country    Electronic    Hip-Hop    Jazz    Rap    Rock
0    0        0        0            0          0             0          1       0      0
1    0        0        0            0          0             0          0       1      0
2    0        0        0            0          1             0          0       0      0
3    0        0        0            0          0             0          0       0      1
4    0        0        0            0          0             0          0       1      0

music_dummies = pd.concat([music_df, music_dummies], axis=1)

music_dummies = music_dummies.drop("genre", axis=1)
Supervised Learning with scikit-learn

Encoding dummy variables

music_dummies = pd.get_dummies(music_df, drop_first=True)

print(music_dummies.columns)

Index(['popularity', 'acousticness', 'danceability', 'duration_ms', 'energy',
       'instrumentalness', 'liveness', 'loudness', 'speechiness', 'tempo',
       'valence', 'genre_Anime', 'genre_Blues', 'genre_Classical',
       'genre_Country', 'genre_Electronic', 'genre_Hip-Hop', 'genre_Jazz',
       'genre_Rap', 'genre_Rock'],
      dtype='object')
Supervised Learning with scikit-learn

Linear regression with dummy variables

from sklearn.model_selection import cross_val_score, KFold
from sklearn.linear_model import LinearRegression
X = music_dummies.drop("popularity", axis=1).values
y = music_dummies["popularity"].values
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, 
                                                    random_state=42)

kf = KFold(n_splits=5, shuffle=True, random_state=42)

linreg = LinearRegression()

linreg_cv = cross_val_score(linreg, X_train, y_train, cv=kf,
                            scoring="neg_mean_squared_error")

print(np.sqrt(-linreg_cv))

[8.15792932, 8.63117538, 7.52275279, 8.6205778, 7.91329988]
Supervised Learning with scikit-learn

Let's practice!

Supervised Learning with scikit-learn

Preparing Video For Download...