Multivariable/Multiple Linear Regression in Scikit Learn?

Question

I have a dataset (dataTrain.csv & dataTest.csv) in .csv file with this format:

Temperature(K),Pressure(ATM),CompressibilityFactor(Z)
273.1,24.675,0.806677258
313.1,24.675,0.888394713
...,...,...

And able to build a regression model and prediction with this code:

import pandas as pd
from sklearn import linear_model

dataTrain = pd.read_csv("dataTrain.csv")
dataTest = pd.read_csv("dataTest.csv")
# print df.head()

x_train = dataTrain['Temperature(K)'].reshape(-1,1)
y_train = dataTrain['CompressibilityFactor(Z)']

x_test = dataTest['Temperature(K)'].reshape(-1,1)
y_test = dataTest['CompressibilityFactor(Z)']

ols = linear_model.LinearRegression()
model = ols.fit(x_train, y_train)

print model.predict(x_test)[0:5]

However, what I want to do is multivariable regression. So, the model will be CompressibilityFactor(Z) = intercept + coef*Temperature(K) + coef*Pressure(ATM)

How to do that in scikit-learn?

Solution

If your code above works for univariate, try this

import pandas as pd
from sklearn import linear_model

dataTrain = pd.read_csv("dataTrain.csv")
dataTest = pd.read_csv("dataTest.csv")
# print df.head()

x_train = dataTrain[['Temperature(K)', 'Pressure(ATM)']].to_numpy().reshape(-1,2)
y_train = dataTrain['CompressibilityFactor(Z)']

x_test = dataTest[['Temperature(K)', 'Pressure(ATM)']].to_numpy().reshape(-1,2)
y_test = dataTest['CompressibilityFactor(Z)']

ols = linear_model.LinearRegression()
model = ols.fit(x_train, y_train)

print model.predict(x_test)[0:5]

Comments

[rtk22]

Just include both Temperature and Pressure in your xtrain, xtest. x_train = dataTrain[["Temperature(K)", "Pressure(ATM)"]] and then the same for x_test.

[Warren O'Neill]

DataFrames don't have a reshape function. To run the above code I have to use values first, eg x_train = dataTrain[['Temperature(K)', 'Pressure(ATM)']].values.reshape(-1,2).