import pandas as pd
import numpy as np

from sklearn.model_selection import train_test_split

df = pd.read_csv('spaceship_passengers.csv')

df.head()

	PassengerId	Name	HomePlanet	CryoSleep	Destination	Age	VIP	RoomService	FoodCourt	ShoppingMall	Spa	VRDeck	CabinDeck	CabinNumber	CabinSide	Transported
0	0001_01	Maham Ofracculy	Europa	False	TRAPPIST-1e	39.0	False	0.0	0.0	0.0	0.0	0.0	B	0.0	P	False
1	0002_01	Juanna Vines	Earth	False	TRAPPIST-1e	24.0	False	109.0	9.0	25.0	549.0	44.0	F	0.0	S	True
2	0003_01	Altark Susent	Europa	False	TRAPPIST-1e	58.0	True	43.0	3576.0	0.0	6715.0	49.0	A	0.0	S	False
3	0003_02	Solam Susent	Europa	False	TRAPPIST-1e	33.0	False	0.0	1283.0	371.0	3329.0	193.0	A	0.0	S	False
4	0004_01	Willy Santantines	Earth	False	TRAPPIST-1e	16.0	False	303.0	70.0	151.0	565.0	2.0	F	1.0	S	True

Data Cleaning

1. How many null/missing values has the column CryoSleep?

df['CryoSleep'].isnull().sum()

217

2. How many null/missing values has the column FoodCourt?

df['FoodCourt'].isnull().sum()

183

3. How many null/missing values has the column PassengerId?

df['PassengerId'].isnull().sum()

0

4. Fill the null/missing values in Destination with the most common choice

df['Destination'].value_counts()

Destination
TRAPPIST-1e      5915
55 Cancri e      1800
PSO J318.5-22     796
Name: count, dtype: int64

df['Destination'].fillna('TRAPPIST-1e' , inplace=True)

/tmp/ipykernel_18/767106531.py:1: FutureWarning: A value is trying to be set on a copy of a DataFrame or Series through chained assignment using an inplace method.
The behavior will change in pandas 3.0. This inplace method will never work because the intermediate object on which we are setting values always behaves as a copy.

For example, when doing 'df[col].method(value, inplace=True)', try using 'df.method({col: value}, inplace=True)' or df[col] = df[col].method(value) instead, to perform the operation inplace on the original object.


  df['Destination'].fillna('TRAPPIST-1e' , inplace=True)

5. Fill the null/missing values in RoomService, FoodCourt, ShoppingMall, Spa and VRDeck with the median

df['RoomService'].fillna(df['RoomService'].median() , inplace=True)
df['FoodCourt'].fillna(df['FoodCourt'].median() , inplace=True)
df['ShoppingMall'].fillna(df['ShoppingMall'].median() , inplace=True)
df['Spa'].fillna(df['Spa'].median() , inplace=True)
df['VRDeck'].fillna(df['VRDeck'].median() , inplace=True)

/tmp/ipykernel_18/1974137612.py:1: FutureWarning: A value is trying to be set on a copy of a DataFrame or Series through chained assignment using an inplace method.
The behavior will change in pandas 3.0. This inplace method will never work because the intermediate object on which we are setting values always behaves as a copy.

For example, when doing 'df[col].method(value, inplace=True)', try using 'df.method({col: value}, inplace=True)' or df[col] = df[col].method(value) instead, to perform the operation inplace on the original object.


  df['RoomService'].fillna(df['RoomService'].median() , inplace=True)
/tmp/ipykernel_18/1974137612.py:2: FutureWarning: A value is trying to be set on a copy of a DataFrame or Series through chained assignment using an inplace method.
The behavior will change in pandas 3.0. This inplace method will never work because the intermediate object on which we are setting values always behaves as a copy.

For example, when doing 'df[col].method(value, inplace=True)', try using 'df.method({col: value}, inplace=True)' or df[col] = df[col].method(value) instead, to perform the operation inplace on the original object.


  df['FoodCourt'].fillna(df['FoodCourt'].median() , inplace=True)
/tmp/ipykernel_18/1974137612.py:3: FutureWarning: A value is trying to be set on a copy of a DataFrame or Series through chained assignment using an inplace method.
The behavior will change in pandas 3.0. This inplace method will never work because the intermediate object on which we are setting values always behaves as a copy.

For example, when doing 'df[col].method(value, inplace=True)', try using 'df.method({col: value}, inplace=True)' or df[col] = df[col].method(value) instead, to perform the operation inplace on the original object.


  df['ShoppingMall'].fillna(df['ShoppingMall'].median() , inplace=True)
/tmp/ipykernel_18/1974137612.py:4: FutureWarning: A value is trying to be set on a copy of a DataFrame or Series through chained assignment using an inplace method.
The behavior will change in pandas 3.0. This inplace method will never work because the intermediate object on which we are setting values always behaves as a copy.

For example, when doing 'df[col].method(value, inplace=True)', try using 'df.method({col: value}, inplace=True)' or df[col] = df[col].method(value) instead, to perform the operation inplace on the original object.


  df['Spa'].fillna(df['Spa'].median() , inplace=True)
/tmp/ipykernel_18/1974137612.py:5: FutureWarning: A value is trying to be set on a copy of a DataFrame or Series through chained assignment using an inplace method.
The behavior will change in pandas 3.0. This inplace method will never work because the intermediate object on which we are setting values always behaves as a copy.

For example, when doing 'df[col].method(value, inplace=True)', try using 'df.method({col: value}, inplace=True)' or df[col] = df[col].method(value) instead, to perform the operation inplace on the original object.


  df['VRDeck'].fillna(df['VRDeck'].median() , inplace=True)

type(df['VIP'][0])

bool

6. Fill the null/missing values in VIP with the most common value

df['VIP'].fillna(False , inplace = True)

/tmp/ipykernel_18/2530139235.py:1: FutureWarning: Downcasting object dtype arrays on .fillna, .ffill, .bfill is deprecated and will change in a future version. Call result.infer_objects(copy=False) instead. To opt-in to the future behavior, set `pd.set_option('future.no_silent_downcasting', True)`
  df['VIP'].fillna(False , inplace = True)

8. Drop the previously defined columns (PassengerId and Name)

df.drop(columns=['Name' , 'PassengerId'], inplace=True)

9. Drop any other row that contains a null value

df.dropna(inplace=True)

Feature Engineering

Encode the features previously identified in its own dataframe

features_to_encode = ['VIP' , 'CryoSleep' , 'HomePlanet' , 'Destination' , 'CabinDeck' , 'CabinSide']

df[features_to_encode].info()

<class 'pandas.core.frame.DataFrame'>
Index: 7919 entries, 0 to 8692
Data columns (total 6 columns):
 #   Column       Non-Null Count  Dtype 
---  ------       --------------  ----- 
 0   VIP          7919 non-null   bool  
 1   CryoSleep    7919 non-null   object
 2   HomePlanet   7919 non-null   object
 3   Destination  7919 non-null   object
 4   CabinDeck    7919 non-null   object
 5   CabinSide    7919 non-null   object
dtypes: bool(1), object(5)
memory usage: 378.9+ KB

df_encoded = pd.get_dummies(df[features_to_encode].astype(str))

Remove the original encoded features from df

df_no_categorical = df.drop(['VIP' , 'CryoSleep' , 'HomePlanet' , 'Destination' , 'CabinDeck' , 'CabinSide'] ,axis=1)

Create a new dataframe combining df_no_categorical and df_encoded

df_final = pd.concat([df_no_categorical,df_encoded],axis=1)

Finally, separate the target variable Transported from the training data

df_train = df_final.drop('Transported' , axis=1)
transported = df_final['Transported']

Model selection and tuning

from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import GridSearchCV

These splits are given and set by us, don't change them:

X_train, X_test, y_train, y_test = train_test_split(df_train, transported, test_size=0.2, random_state=0)

---

Random Forest

Use a GridSearchCV to find the best max_depth parameter for a RandomForestClassifier

Given the RandomForestClassifier created (with random_state=42, important, don't change it!) instantiate a GridSearchCV to find the best possible parameter for max_depth, in the range 5 to 25.

rf = RandomForestClassifier(random_state=42)

# your grid of parameters
param_grid = {'max_depth':range(5,26)}

grid_search = GridSearchCV(rf, param_grid, cv=5)

grid_search.fit(X_train, y_train)

GridSearchCV(cv=5, estimator=RandomForestClassifier(random_state=42),
             param_grid={'max_depth': range(5, 26)})

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

What's the best hyperparameter value for max_depth?

grid_search.best_params_

{'max_depth': 20}

Create a RandomForestClassifier that achieves at least 0.75 in precision and at least 0.80 in recall

model = RandomForestClassifier(random_state=42,max_depth=20)

model.fit(X_train, y_train)

RandomForestClassifier(max_depth=20, random_state=42)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

y_pred = model.predict(X_test)

print(f"Model precision: {precision_score(y_test, y_pred)}")
print(f"Model recall: {recall_score(y_test, y_pred)}")

Model precision: 0.8462549277266754
Model recall: 0.7777777777777778

from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay , recall_score,precision_score

cm = confusion_matrix(y_test, y_pred)
ConfusionMatrixDisplay(cm).plot()

<sklearn.metrics._plot.confusion_matrix.ConfusionMatrixDisplay at 0x7070fae436d0>