R^2 score is different in sklearn than MS-Excel RSQ() function.

[rohan472000]

I'm trying to find R^2 score for my dataframe through sklearn.metrics.r2_score which is giving me different score than Excel (by using RSQ() function), I also noticed that sklearn's R^2 score is lower than excel's R^2 score.

    y_true = [200,280,320,346.47]
    y_pred = [228.6,307.8,414.4,376.5]
    print(r2_score(y_true,y_pred))   ## Output: 0.07

but in ms-excel(by using RSQ() function) it is giving 0.86.

I can't show my data, so I have given a sample data above.
Below is my actual output result, where R^2 score is going beyond 0 to1 .

          a = r2_score(df['PM2.5 (μg/m3)'],df['ConcRT(ug/m3)'])
          b = r2_score(df['PM2.5 (μg/m3)'],df['PM2.5'])
          c = r2_score(df['ConcRT(ug/m3)'],df['PM2.5'])
          print("a:",a)
          print("b:",b)
          print("c:",c)
                
        Output: 
        a: -0.780401801591601
        b: -1.8405566667749902
        c: -1.3988707912193044

[glemaitre]

The definition used in scikit-learn is documented here: https://scikit-learn.org/dev/modules/model_evaluation.html#r2-score

[rohan472000]

so basically RSQ() function in excel gives Pearson correlation coefficient whereas sklearn.metrics.r2_score gives coefficients of determination.