Styjun

I have a CSV of twitter profile data, containing: name, description, followers count, following count, bot (class I want to predict)

I have successfully executed a classification model when using just the CountVectorizer values (xtrain) and Bot (ytrain). But have not been able to add this feature to my set of other features.

vectorizer = CountVectorizer()
CountVecTest = vectorizer.fit_transform(training_data.description.values.astype('U'))
CountVecTest = CountVecTest.toarray()
arr = sparse.coo_matrix(CountVecTest)
training_data["NewCol"] = arr.toarray().tolist()

rf = RandomForestClassifier(criterion='entropy', min_samples_leaf=10, min_samples_split=20)
rf = rf.fit(training_data[["followers_count","friends_count","NewCol","bot"]], training_data.bot)

ERROR:

---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
<ipython-input-54-7d67a6586592> in <module>()
 1 rf = RandomForestClassifier(criterion='entropy', min_samples_leaf=10, min_samples_split=20)
----> 2 rf = rf.fit(training_data[["followers_count","friends_count","NewCol","bot"]], training_data.bot)

D:_MyFiles_LibrariesDocumentsAnaconda3libsite-packagessklearnensembleforest.py in fit(self, X, y, sample_weight)
 245 """
 246 # Validate or convert input data
--> 247 X = check_array(X, accept_sparse="csc", dtype=DTYPE)
 248 y = check_array(y, accept_sparse='csc', ensure_2d=False, dtype=None)
 249 if sample_weight is not None:

D:_MyFiles_LibrariesDocumentsAnaconda3libsite-packagessklearnutilsvalidation.py in check_array(array, accept_sparse, dtype, order, copy, force_all_finite, ensure_2d, allow_nd, ensure_min_samples, ensure_min_features, warn_on_dtype, estimator)
 431 force_all_finite)
 432 else:
--> 433 array = np.array(array, dtype=dtype, order=order, copy=copy)
 434 
 435 if ensure_2d:

ValueError: setting an array element with a sequence.

I did some debugging:

print(type(training_data.NewCol))
print(type(training_data.NewCol[0]))
>>> <class 'pandas.core.series.Series'>
>>> <class 'numpy.ndarray'>

Any help would be appreciated.

I would do this the other way around and add your features to your vectorization. Here is what I mean with a toy example:

from sklearn.feature_extraction.text import CountVectorizer
from sklearn.ensemble import RandomForestClassifier
import pandas as pd
import numpy as np
from scipy.sparse import hstack, csr_matrix

Suppose now you have you features in a dataframe called df and your labels in y_train:

df = pd.DataFrame("a":[1,2],"b":[2,3],"c":['we love cars', 'we love cakes'])
y_train = np.array([0,1])

You want to perform a text vectorization on column c and add the features a and b to your vectorization.

vectorizer = CountVectorizer()
CountVecTest = vectorizer.fit_transform(df.c)

CountVecTest.toarray()

This will return:

array([[0, 1, 1, 1],
 [1, 0, 1, 1]], dtype=int64)

But CountVecTest now is a scipy sparse matrix. So what you need to do is add your features to this matrix. Like this:

X_train = hstack([CountVecTest, csr_matrix(df[['a','b']])])

X_train.toarray()

This will return, as expected:

array([[0, 1, 1, 1, 1, 2],
 [1, 0, 1, 1, 2, 3]], dtype=int64)

Then you can train your random forest:

rf = RandomForestClassifier(criterion='entropy', min_samples_leaf=10, min_samples_split=20)
rf.fit(X_train, y_train)

NB: In the code snippet you provided, you passed the label info (the "bot" column) to the training features, which you should obviously not do.