CODEWITHSUNDEEP
python-3.xpyspark
Statmonger
Statmonger

Reputation: 425

AttributeError: 'str' object has no attribute 'fit' - Pyspark

I am trying to run the script below in PySpark3 and receiving the error message that follows. I am using this has something to do with formatting but I am not sure how to go about doing so. Any help would be much appreciated. 

train,test = df.randomSplit([0.7,0.3])

models = ["LinearRegression()","DecisionTreeRegressor()","RandomForestRegressor()","GBTRegressor()"]

for model in models:

    # Fit our model
    M = model
    fitModel = M.fit(train)

    # Load the Summary
    trainingSummary = fitModel.summary

#     trainingSummary.residuals.show()
    print("Training RMSE: %f" % trainingSummary.rootMeanSquaredError)
    print("Training r2: %f" % trainingSummary.r2)

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-15-d0b941a7170e> in <module>()
      8     # Fit our model
      9     M = model
---> 10     fitModel = M.fit(train)
     11 
     12     # Load the Summary

AttributeError: 'str' object has no attribute 'fit'

Upvotes: 0

Views: 1751

Answers (2)

Statmonger
Statmonger

Reputation: 425

I think this way is actually more efficient....

This way you can still iterate through a list. 

def ClassTrainEval(model):

    fitModel = model.fit(train)

    # Load the Summary
    trainingSummary = fitModel.summary

    print("Training RMSE: %f" % trainingSummary.rootMeanSquaredError)
    print("Training r2: %f" % trainingSummary.r2)

models = [LogisticRegression(),NaiveBayes(),OneVsRest(),LinearSVC()] 

for model in models:
    ClassTrainEval(classifier)

Upvotes: 1

Duy Nguyen
Duy Nguyen

Reputation: 1005

You have to modify your models

models = [LinearRegression, DecisionTreeRegressor, RandomForestRegressor, GBTRegressor]

Because with your current define, they are definitely a string. Better way is to add modules inside list and instance it every loop.

To be precies models = ["LinearRegression()"] the element inside is String, not an object which doesn't have any method.

Upvotes: 1

Related Questions