Change column value based on calculated mean of other column in python pandas - python-3.x

Question

I am newbie to pandas. I have gone through many questions, but doesn't found an answer.
I have the following datasets.
Name || Price || Cuisine Category || City || Region || Cuisine Types || Rating Types || Rating
Pizza || 600 || Fast Food,Pizza || Ajmer || Ana Saga || Quick Bites || Good || 3.9
... ... ... ... ... ... ... ... ...
Chawla's || 300 || Beverages || Ajmer || Sagar Lake || Cafe || Average || 3.3
Masala || 0 || North,South Indian || Ajmer || Ram Ganj || Mess || None || NEW
I want to change the value of:
Rating where it is a NEW, on the basis of Average Rating of that particular Cuisine Types and then Rating Types based on the calculated Ratings
Price where it is a 0, on the basis of Average Price of that particular Region
My try for changing price:
Reading CSV File
data = pd.read_csv('/content/Ajmer.csv')
calculate Region wise mean of Price
gregion = round(data.groupby('Region')['Price'].mean())
Trying to replace 0 of price column
data['Price'] = data['Price'].replace(0, gregion[data['Region']])
But my price column is unchanged.
My try for changing Rating:
Reading CSV file
data2 = pd.read_csv('/content/Ajmer.csv')
Creating separate data frame, so that it wont affect mean value.
filtered_rating = data2[(data2['Rating'] == 'NEW') | (data2['Rating'] == '-') | (data2['Rating'] == 'Opening')]
Dropping from original data2
data2.drop(data2.loc[data['Rating']=='NEW'].index, inplace=True)
data2.drop(data2.loc[data['Rating']=='-'].index, inplace=True)
data2.drop(data2.loc[data['Rating']=='Opening'].index, inplace=True)
Calculating Cuisine Types wise mean of Rating
c = round(data2.groupby('Cuisine Types')['Rating'].mean(),1)
which gives me output as follows:
Cuisine Types
Bakery 3.4
Confectionery 3.4
Dessert Parlor 3.5
...
Quick Bites 3.4
Sweet Shop 3.4
Name: Rating, dtype: float64
Trying to replace values
filtered_rating['Rating'].replace('NEW', c[data2['Region']], inplace=True)
filtered_rating['Rating'].replace('-', c[data2['Region']], inplace=True)
filtered_rating['Rating'].replace('Opening', c[data2['Region']], inplace=True)
But my Rating column is unchanged.
Expected output
Mean Price of that particular Region of row where the price is zero in Price column
Mean Rating of that particular Cuisine Type of row where the Rating is NEW in Rating column
Can anyone help me out to do this?
Thanks in Advance!
I will be more than happy for your help

Answer 1

lets say you have a data like following.
data
name region price cuisine_type rating_type rating
0 pizza NY 500 fast food average 3.3
1 burger NY 350 fast food good 4.1
2 lobster LA 1500 seafood good 4.5
3 mussels LA 1000 seafood average 3.9
4 shawarma NY 300 mediterranean average 3.4
5 kabab LA 600 mediterranean good 4
6 pancake NY 250 breakfast average 3.7
7 waffle LA 450 breakfast good 4.2
8 fries NY 0 fast food None NEW
9 crab LA 0 seafood None Opening
10 tuna sandwich NY 0 seafood None NEW
11 onion rings LA 0 fast food None Opening
Now according to your question we need to replace the rating when it is NEW or Opening with mean rating of respective cuisine_type. And price when its 0 with mean price of respective region. And update rating type for None at the end.
#get a list of cuisine types
cuisine_type_list=data.cuisine_type.unique().tolist()
cuisine_type_list
['fast food', 'seafood', 'mediterranean', 'breakfast']
#get a list of regions
region_list=data.region.unique().tolist()
region_list
['NY', 'LA']
#replace the ratings
for i in cuisine_type_list:
data.loc[(data.cuisine_type==i) & (data.rating.isin(['NEW', 'Opening'])), 'rating']=round(data.loc[(data.cuisine_type==i) & (data.rating.isin(['NEW', 'Opening'])==False)].rating.mean(), 2)
#replace price when 0
for i in region_list:
data.loc[(data.region==i) & (data.price==0), 'price']=round(data.loc[(data.region==i) & (data.price!=0)].price.mean(), 2)
#function to assign rating type (assuming good for rating>=4)
def calculate_rating_type(row):
if row['rating'] >= 4:
return 'good'
else:
return 'average'
#update rating type
data.loc[data.rating_type.isnull(), 'rating_type']=data.loc[data.rating_type.isnull()].apply(lambda row: calculate_rating_type(row), axis=1)
this is the data after updating
data
name region price cuisine_type rating_type rating
0 pizza NY 500 fast food average 3.3
1 burger NY 350 fast food good 4.1
2 lobster LA 1500 seafood good 4.5
3 mussels LA 1000 seafood average 3.9
4 shawarma NY 300 mediterranean average 3.4
5 kabab LA 600 mediterranean good 4
6 pancake NY 250 breakfast average 3.7
7 waffle LA 450 breakfast good 4.2
8 fries NY 350 fast food average 3.7
9 crab LA 887.5 seafood good 4.2
10 tuna sandwich NY 350 seafood good 4.2
11 onion rings LA 887.5 fast food average 3.7

Answer 2

You can try the following code:
gregion = round(data.groupby('Region')['Price'].mean())
# convert your group by to DataFrame
gregion = pd.DataFrame(gregion)
gregion.reset_index(inplace=True)
# merge the datas and drop the new column that is created
data = data.merge(gregion, left_on='Region', right_on='Region', suffixes=('_x', ''))
data = data.drop(columns={'Price_x'})
filtered_rating = data[(data['Rating'] == 'NEW') | (data['Rating'] == '-') | (data['Rating'] == 'Opening')]
# you don't need to re-upload the file
data2 = data.copy()
data2.drop(data2.loc[data2['Rating']=='NEW'].index, inplace=True)
data2.drop(data2.loc[data2['Rating']=='-'].index, inplace=True)
data2.drop(data2.loc[data['Rating']=='Opening'].index, inplace=True)
# do the same with c
c = round(data2.groupby('Cuisine Types')['Rating'].mean(),1)
c = pd.DataFrame(c)
c.reset_index(inplace=True)
filtered_rating = filtered_rating.merge(c, left_on='Cuisine Types', right_on='Cuisine Types', how='left', suffixes=('_x', ''))
filtered_rating = filtered_rating.drop(columns={'Rating_x'})
Hope this helps.