Import Libraries and File

In [1]:
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
In [2]:
df = pd.read_csv('GBBO.csv')

Initial Inspection of the dataset

In [3]:
print(df.head())

     Airdate  Season  Episode    Theme             Signature  \
0  8/17/2010       1        1     Cake                  Cake   
1  8/24/2010       1        2  Biscuit  Personality Biscuits   
2  8/31/2010       1        3    Bread       Signature Bread   
3   9/7/2010       1        4  Pudding       Classic Pudding   
4  9/14/2010       1        5   Pastry           Savoury Pie   

   Signature Time (min)                Technical  Technical Time (min)  \
0                 180.0        Victoria Sandwich                   NaN   
1                 120.0                   Scones                  60.0   
2                 210.0                      Cob                 150.0   
3                 150.0  Mini Hot Lemon Soufflés                  40.0   
4                 150.0          Cornish Pasties                  90.0   

                                 Showstopper  Showstopper Time (min)  \
0                 Chocolate Celebration Cake                     NaN   
1  Meringue, Choux, and Macaron Petits Fours                   240.0   
2              12 Sweet and 12 Savoury Rolls                   360.0   
3            Crumble, Bread, & Suet Puddings                   300.0   
4          Savory Canapés and Sweet Tartlets                   300.0   

   MyRating (out of 10)  MyViewership  
0                   8.0          8.57  
1                   7.8         11.14  
2                   7.9         11.47  
3                   7.7         11.03  
4                   7.8         11.93
In [4]:
print(df.info())

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 134 entries, 0 to 133
Data columns (total 12 columns):
 #   Column                  Non-Null Count  Dtype  
---  ------                  --------------  -----  
 0   Airdate                 134 non-null    object 
 1   Season                  134 non-null    int64  
 2   Episode                 134 non-null    int64  
 3   Theme                   134 non-null    object 
 4   Signature               134 non-null    object 
 5   Signature Time (min)    124 non-null    float64
 6   Technical               133 non-null    object 
 7   Technical Time (min)    122 non-null    float64
 8   Showstopper             134 non-null    object 
 9   Showstopper Time (min)  117 non-null    float64
 10  MyRating (out of 10)    134 non-null    float64
 11  MyViewership            134 non-null    float64
dtypes: float64(5), int64(2), object(5)
memory usage: 12.7+ KB
None

The frist row contains a NaN value, so let's explore further:

In [5]:
missing_df = df.isnull().sum().to_frame().reset_index().rename({"index" : 'Column', 0: 'Missing Values'}, axis =1)
display(missing_df)


# This could have also worked... showing the proportion rather than the counts
# missing_df = (df.isnull().sum() / (len(df.index)) * 100).to_frame().reset_index().rename({"index" : 'Column',0:'Proportion'}, axis = 1)
# display(missing_df)
Column Missing Values
0 Airdate 0
1 Season 0
2 Episode 0
3 Theme 0
4 Signature 0
5 Signature Time (min) 10
6 Technical 1
7 Technical Time (min) 12
8 Showstopper 0
9 Showstopper Time (min) 17
10 MyRating (out of 10) 0
11 MyViewership 0

The Technical field is just a text description so we can ignore the one missing value. Also, the technical time and Showstopper time are very specific to what was made, so we'll likely have to ignore them too.

However, the Signature time is likely related to the Theme. Let's aggregative the Signature Times with the Theme field and fill in blanks using these averages.

In [ ]:
In [6]:
category_means = df.groupby('Theme')['Signature Time (min)'].mean()


for category, mean_value in category_means.items():
    df.loc[(df['Theme'] == category) & (df['Signature Time (min)'].isnull()), 'Signature Time (min)'] = mean_value
In [7]:
# Another check:

missing_df = df.isnull().sum().to_frame().reset_index().rename({"index" : 'Column', 0: 'Missing Values'}, axis =1)
display(missing_df)
Column Missing Values
0 Airdate 0
1 Season 0
2 Episode 0
3 Theme 0
4 Signature 0
5 Signature Time (min) 4
6 Technical 1
7 Technical Time (min) 12
8 Showstopper 0
9 Showstopper Time (min) 17
10 MyRating (out of 10) 0
11 MyViewership 0
In [8]:
missing_values_rows = df[df['Signature Time (min)'].isnull()]

print(missing_values_rows['Theme'])

87        Dairy
90     Festival
100    The '80s
125    Biscuits
Name: Theme, dtype: object
In [9]:
df.dropna(subset=['Signature Time (min)'], inplace=True)