[FIXED] Why .fit() needs 2D array as the first parameter?

Issue

import pandas as pd
import numpy as np
from sklearn import linear_model
import matplotlib.pyplot as plt

df = pd.read_csv('homeprices.csv')

plt.xlabel('area')
plt.ylabel('price')
plt.scatter(df.area,df.price,color='red',marker='.')


reg = linear_model.LinearRegression()
reg.fit(df.area,df.price)

Error Message:

ValueError: Expected 2D array, got 1D array instead: array=[2600 3000 3200 3600 4000]. Reshape your data either using array.reshape(-1, 1) if your data has a single feature or array.reshape(1, -1) if it contains a single sample.

It works fine if I write it as :

reg.fit(df[['area']],df.price)

I would like to know the reason behind it because The second argument is passed as df.price.

My csv file:

• area,price
• 2600,550000
• 3000,565000
• 3200,610000
• 3600,680000
• 4000,725000

Solution

From the documentation, variable x should be declared as

X{array-like, sparse matrix} of shape (n_samples, n_features)

When you declare:

• x = df.area or x = df['area'] the x will become Series type with the size (n,). The size should be (n, z), where z can be any positive integer.

• x = df[['area']] the x will become DataFrame type with the size (5, 1) which makes an x an acceptable input.

• y = df.price the y will become Series type with the size (5,) which s acceptable input.

y: array-like of shape (n_samples,)

But if I were you I declare x and y as:

x = [[i] for i in df['area']]
y = [i for i in df['price']]

which makes both x and y as the list structure and set the size to the (5, 1), so in the future if you want to run in any ML library (tensorflow, pytorch, keras, ...) you won't have any difficulties.

Answered By - Ahx