单一数据可通过具有多个分类特征的线性回归模型进行预测

Question

I built a linear regression model to predict the sales numbers for a product, In my case I have 5 features, 4 of them are categorical.

MONTH REGION INTERVENANT CONFIG WEIGHT SALES_NB

I used OneHotEncoder

from sklearn.preprocessing import OneHotEncoder
onehotencoder = OneHotEncoder(categorical_features = [0,1,2,3])
X = onehotencoder.fit_transform(X).toarray()
X = X [:, 1:]

(correct me if I am wrong)

I want to know how do I format my data to pass it to predict(). Actually if I pass:

Xnew = np.array([[2,2,14895,614,0.1]])
ynew = regressor.predict(Xnew)

I got this error:

ValueError: shapes (1,4) and (428,) not aligned: 4 (dim 1) != 428 (dim 0)

Answer 1

Try encoding the new sample with onehotencoder before you pass it to the predictor:

Xnew = np.array([[2,2,14895,614,0.1]])
Xnew_encoded = onehotencoder.transform(Xnew)
ynew = regressor.predict(Xnew_encoded)