I am trying to make a regression model in order to predict ratings (1-5) based on words that appear (the regression doesn't have to perform well per se, it's more about the methodology applied). I created a term frequency matrix with this code:
bow = df.Review2.str.split().apply(pd.Series.value_counts)
which look like this:
I am now interested in deleting columns (words) that rarely appear throughout the reviews. Moreover, I want to iterate through only the reviews (rows) that have a Rating
value which is not NaN
.
here is my attempt:
# Delete row if Rating less than 1
for index, row in df.iterrows():
if (df.Rating[index] < 1):
bow.drop(bow.index[index], axis=0, inplace = True)
# Delete column if word occurs less than 50 times
sum1 = bow.sum(axis=0)
cntr = 0
for i in sum1:
if (i < 50):
bow.drop(bow.index[cntr], axis=1, inplace = True)
cntr += 1
This doesn't seem to do the work as it leaves words that occur only once.
EDIT:
This is my sparse dataframe containing occurrences of words. Col -> words; Rows -> sentences (item's reviews) (I have 1.5k items, thus 1.5k rows)
hi this are just some random words I don t ... zing zingy zingzang
0 1.0 NaN 1.0 1.0 1.0 NaN NaN NaN NaN NaN ... NaN NaN NaN
1 NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN
2 NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN
3 NaN NaN NaN NaN NaN NaN NaN NaN NaN NaN ... NaN NaN NaN
4 NaN NaN NaN NaN NaN NaN NaN NaN NaN 1.0 ... NaN NaN NaN
Rating
is a single column of my original dataframe containing integers in the [1,5]
range or NaN
You can use vectorised operations instead of manual iteration:
# filter out rows where Rating < 1
bow = bow[~(bow['Rating'] < 1)]
# filter out columns where sum < 50
bow = bow.loc[:, ~(bow.sum(0) < 50)]
Or simultaneously:
# filter rows and columns with Boolean series
bow = bow.loc[~(bow['Rating'] < 1), ~(bow.sum(0) < 50)]
I made this working toy example:
import pandas as pd
import numpy as np
# Create a toy daframe
df = pd.DataFrame(np.arange(12).reshape(3,4),columns=['A', 'B', 'C', 'D'])
print(df)
# A B C D
#-------------
# 0 0 1 2 3
# 1 4 5 6 7
# 2 8 9 10 11
# Sum all the values for each column
column_sum = df.sum(axis=0)
print(column_sum)
# A 12
# B 15
# C 18
# D 21
# Iterate over Columns name and sum value
for key,value in zip(df.keys(),sum1):
if(value < 16):
df.drop(columns=key, axis=1, inplace = True)
print(df)
# C D
# 0 2 3
# 1 6 7
# 2 10 11
so I guess that if you change your code to:
for key,value in zip(df.keys(),sum1):
if(value < 50):
bow.drop(columns=key, axis=1, inplace = True)
it should get the job done.
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.