Encoding error using altair interactive plot?

Question

I'm trying to use the following data df_roc to plot an ROC curve using altair:

    Threshold   TPR     FPR
0   0.1     1.000000    0.941176
1   0.2     1.000000    0.705882
2   0.3     0.923077    0.588235
3   0.4     0.846154    0.470588
4   0.5     0.692308    0.352941
5   0.6     0.615385    0.235294
6   0.7     0.461538    0.117647
7   0.8     0.307692    0.058824
8   0.9     0.076923    0.000000

This is the code I'm attempting to use in order to make the interactive plot:

base = alt.Chart(df_roc, 
                 title='ROC Curve of KNN'
                ).properties(width=300)

roc_curve = base.mark_line(point=True).encode(
    alt.X('fpr', title='False Positive Rate (FPR)',  sort=None),
     alt.Y('tpr', title='True Positive Rate (TPR) (a.k.a Recall)'),
)

roc_rule = base.mark_line(color='green').encode(
    x='fpr',
    y='fpr',
    size=alt.value(2)
)


(roc_curve + roc_rule).interactive()

And this is the error I'm getting:

ValueError: fpr encoding field is specified without a type; the type cannot be inferred because it does not match any column in the data.

alt.Chart(...)

I've tried to google around a bit and try some information on it, but there wasn't really much. Has anyone come across a solve for this or help me to find a way around it?

I would really prefer to be able to use altair for this as opposed to other plotting packages.

Can anyone help me?

Answer 1

Column names in Altair (and in pandas in general) are case-sensitive. It appears that your data has columns named "TPR" and "FPR' , but your chart specifies columns named "tpr" and "fpr" .

Change the case and your chart should work:

base = alt.Chart(df_roc, 
                 title='ROC Curve of KNN'
                ).properties(width=300)

roc_curve = base.mark_line(point=True).encode(
    alt.X('FPR', title='False Positive Rate (FPR)',  sort=None),
     alt.Y('TPR', title='True Positive Rate (TPR) (a.k.a Recall)'),
)

roc_rule = base.mark_line(color='green').encode(
    x='FPR',
    y='TPR',
    size=alt.value(2)
)


(roc_curve + roc_rule).interactive()