How do I fix an Index Error in Numpy for use in a model for machine learning?

Question

I was trying to predict the probability of a person to be using a certain drug. One of the vital predictions is showing an IndexError.

This is for a prediction. I've used this structure with no problems on different predictions with no problems.

import numpy as np
import pandas as pd
import sklearn.ensemble as skle
drug = pd.read_sas('C:/Users/hamee/Downloads/DUQ_I.xpt')
mod = skle.RandomForestClassifier()
fitmod = mod.fit(drug2[["DUQ200", "DUQ240", "DUQ250", "DUQ290", "DUQ330", "DUQ370"]], drug2["DUQ240"])
Pred = fitmod.predict_proba(drug2[["DUQ200", "DUQ240", "DUQ250", "DUQ290", "DUQ330", "DUQ370"]])
Brier = np.mean((Pred[:,1]-drug2["DUQ290"]**2))

I expected an output of a decimal or a large number, The output was :

IndexError                                Traceback (most recent call last)
<ipython-input-19-90c24bde1c32> in <module>
----> 1 Brier = np.mean((Pred[:,1]-drug2["DUQ290"]**2))

IndexError: index 1 is out of bounds for axis 1 with size 1

Right before that line (`Brier = ...`) print `Pred.shape`. What the result? — wwii, Apr 27 '19 at 02:46
What is the shape of `drug2["DUQ290"]`?? Try `Pred[:,0]-...` — wwii, Apr 27 '19 at 02:51
I just tried the "Pred[:,0], which worked. Thank you for your help — Hameed Abdul, Apr 27 '19 at 23:49
[Can I answer my own question?](https://stackoverflow.com/help/self-answer) — wwii, Apr 28 '19 at 02:05

Rocky Li · Accepted Answer · 2019-04-28T03:59:36.683

1

Assuming everything up to your Pred is working correctly

Your Pred is a one dimensional array, try Pred.shape, it should return a tuple of (length, ). Hence you only need to do

Brier = np.mean((Pred-drug2["DUQ290"]**2))

Update: since your Pred.shape is (539, 1), it should be:

Brier = np.mean((Pred[:,0]-drug2["DUQ290"]**2))

as all list indices starts from 0.

edited Apr 28 '19 at 03:59

answered Apr 27 '19 at 02:17

Rocky Li

5,641
2
17
33

The data is not a one dimensional array, this is the output after this change: – Hameed Abdul Apr 27 '19 at 23:46
--------------------------------------------------------------------------- Exception Traceback (most recent call last) in ----> 1 Brier = np.mean((Pred-drug2["DUQ290"]**2)) ~\Anaconda3\ Exception: Data must be 1-dimensional – Hameed Abdul Apr 27 '19 at 23:46
So what is `Pred.shape`? – Rocky Li Apr 27 '19 at 23:49
Pred.shape = (539,1) – Hameed Abdul Apr 28 '19 at 01:53
The correct way to address this would then be: `Pred[:, 0]` – Rocky Li Apr 28 '19 at 03:58

How do I fix an Index Error in Numpy for use in a model for machine learning?

I expected an output of a decimal or a large number, The output was :

1 Answers1