sklearn: Found input variables with inconsistent numbers of samples: [1, 99]

Question

I'm trying to build a simple regression line with pandas in spyder. After executing the following code, I got this error:

Found input variables with inconsistent numbers of samples: [1, 99]

the code:

import numpy as np
import pandas as pd

dataset = pd.read_csv('Phil.csv')

x = dataset.iloc[:, 0].values
y = dataset.iloc[:, 2].values

from sklearn.linear_model import LinearRegression
regressor = LinearRegression()
regressor.fit(x, y)

I think I know what is the problem, but I'm not quite sure how to deal with the syntax. In the variable explorer, the size of x (and y) is (99L,), and from what I remember it can't be a vector, and it must be size (99,1). same thing for y.

Saw a bunch of related topics, but none of them helped.

Novanovaculite

y can just be (99,) (need not be of shape (99,1)), but X must be a 2-d shape. Try x = x.reshape(-1,1) before fitting. – Exodontist

Answer 1

Referring to the sklearn documentation for LinearRegression (http://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LinearRegression.html#sklearn.linear_model.LinearRegression.fit), the X vector needs to conform to the specification [n_samples,n_features].

Since you have only a single feature with many samples, the shape should be (99,1) - e.g., a single value per "row" with a single "column".

There are many ways to accomplish this (ref: Efficient way to add a singleton dimension to a NumPy vector so that slice assignments work), in your case, the following should work:

regressor.fit(x[:, None], y)

Don't forget that predict requires the same shape to the data!

Nickinickie

Answer 2

I got a similar issue as well.

ValueError: Found input variables with inconsistent numbers of samples: [20, 10]

I found a solution though. For my case the order of the splitting was not correct

I did

X_train, X_test, y_test, y_train = train_test_split(X,y,test_size=1/3, random_state=0) 

instead of :

X_train, X_test, y_train, y_test = train_test_split(X,y,test_size=1/3, random_state=0) 

Hope it helps future coders who run into similar errors.

Baptista