Data-Visualisation-Challenge-codes-python

import numpy as np import pandas as pd import matplotlib.pyplot as plt import seaborn as sns df=pd.read_csv(r"C:\Users\abhin\Jupyter notebook\ieee datasets\udemy_courses.csv") df.head()

Data Cleaning

df.shape df.isnull().sum() df.info()

filtering rows which have empty values

df[df['instructor_names'].isnull()]

will see this again after univariate analysis on numerical columns

checking for duplicates

duplicates = df[df.duplicated()] print(duplicates)

As we see that there are no duplicates in the dataset

univariate analysis on numerical columns

Univariate on 'rating' df['rating'].describe() df['rating'].plot(kind='kde') df['rating'].plot(kind='hist') plt.xlabel('rating') sns.violinplot(df['rating']) df['rating'].plot(kind='box')

filtering rows with rating between 3 and 0.5

df[(df['rating']<=3) & (0.5<=df['rating'])] Univariate on 'num_subscribers' df['num_subscribers'].describe() ax = df['num_subscribers'].plot(kind='hist', bins=100)

ax.set_xlim(0, 250000) # there are some courses which have subscribers more than 250000 but we are not

seeing them in this plot to better visualize and as more than 99.5% coures have subscribers in this range

ax.get_xaxis().set_major_formatter(plt.FuncFormatter(lambda x, _: f'{int(x):,}'))

plt.xlabel('Number of Subscribers') plt.ylabel('courses count')
plt.title('Histogram of Number of Subscribers') # Add a title plt.show()

courses with subscribers more than 250000

df[df['num_subscribers']>250000] ax = df['num_subscribers'].plot(kind='kde')

ax.set_xlim(0, 40000)

ax.get_xaxis().set_major_formatter(plt.FuncFormatter(lambda x, _: f'{int(x):,}'))

plt.xlabel('Number of Subscribers') plt.ylabel('Frequency')
plt.title('num_subscribers') plt.show() df['num_subscribers'].plot(kind='box') univariate on 'num_reviews' df['num_reviews'].describe() df['num_reviews'].plot(kind='box') ax=df['num_reviews'].plot(kind='kde') ax.set_xlim(0,150000) # here the courses we are seeing have reviews less than 150000

plt.xlabel('num_reviews') plt.ylabel('courses_count_probability') plt.title('num_reviews')

plt.show() ax=df['num_reviews'].plot(kind='hist',bins=10000) ax.set_xlim(0,500) #i am not seeing here courses with reviews more than 500 they are outliers and will see #them after plt.xlabel('num_reviews') plt.ylabel('courses_count') plt.title('reviews')

plt.show()

univariate on categorical columns

Univariate on feature 'category' df['category'].value_counts() ax = df['category'].value_counts().plot(kind='pie', autopct='%1.1f%%')

plt.legend(title='Categories', bbox_to_anchor=(1.05, 1), loc='upper left') plt.ylabel('')
plt.title('Category Distribution')
plt.tight_layout()
plt.show()

df['category'].value_counts().plot(kind='bar') Univariate on feature 'is_paid' df['is_paid'].value_counts() df['is_paid'].value_counts().plot(kind='pie', autopct='%1.1f%%') plt.title('is_paid') plt.ylabel('') plt.tight_layout() plt.show() Univariate on 'instructional_level' df['instructional_level'].value_counts() df['instructional_level'].value_counts(normalize=True)*100 df['instructional_level'].value_counts().plot(kind='pie', autopct='%1.1f%%') plt.ylabel('')
plt.show() #filtering the rows with missing values df[df['instructor_names'].isnull()]

as we can see the numerical features of those 2 rows and compare with other rows we compare

#with other rows so for 1st row if we see marketing courses with num_subscribers and rating around that value as-- fil1=(df['num_subscribers']<=10500) & (df['num_subscribers']>=10000) fil2=(df['rating']<=4.2) & (df['rating']>=3.8) df[(df['category']=='Marketing') & fil1 & fil2] film1=(df['num_subscribers']<=250) & (df['num_subscribers']>=200) film2=(df['rating']<=3.9) & (df['rating']>=3.6) df[(df['category']=='Lifestyle') & film1 & film2]

So we can see for both courses there are enough courses with values in that range so

they don't affect much So i am dropping the 2 courses

df.dropna(inplace=True)

Bivariate on numerical-numerical

import matplotlib.ticker as mticker df[['num_subscribers','rating','num_reviews']].corr() ax=sns.scatterplot(data=df,y='num_subscribers',x='rating') ax.set_ylim(0, 350000)

ax.get_xaxis().set_major_formatter(plt.FuncFormatter(lambda x, _: f'{int(x):,}'))

plt.show() ax=sns.kdeplot(data=df,x='rating', y='num_subscribers') plt.title("subscribers vs rating") ax.set_ylim(0, 100000)

ax.yaxis.set_major_formatter(mticker.FuncFormatter(lambda x, _: f'{int(x):,}'))

plt.xlabel("rating") plt.ylabel("subscribers") plt.show()

bivariate on categorical- categorical

pd.crosstab(df['is_paid'],df['category']) pd.crosstab(df['is_paid'],df['instructional_level'],normalize='columns')*100 pd.crosstab(df['instructional_level'],df['category'],normalize='columns')*100

Bivariate on numerical- categorical

sns.barplot(x='instructional_level',y='num_subscribers',data=df,errorbar=None,estimator='mean') plt.ylabel('avgNum_of_subscribers_forthe_level') with pd.option_context('display.float_format', '{:.2f}'.format): print(df[df['instructional_level']=='All Levels']['num_subscribers'].describe()) sns.barplot(x='category',y='num_subscribers',data=df,errorbar=None,estimator='mean') plt.ylabel('avg subscribers per category') plt.xticks(rotation=90) plt.show() categories_top = df['category'].unique()[:6] categories_bottom = df['category'].unique()[6:]

Create subplots

fig, axes = plt.subplots(2, 1, figsize=(12, 12))

First subplot (Top categories)

sns.violinplot( data=df[df['category'].isin(categories_top)], x='category', y='num_subscribers', ax=axes[0] ) axes[0].set_ylim(0, 50000) axes[0].get_yaxis().set_major_formatter(plt.FuncFormatter(lambda x, _: f'{int(x):,}'))

axes[0].set_title('Subscribers vs Category (Top Categories)') axes[0].set_ylabel('Number of Subscribers') axes[0].set_xlabel('Category') axes[0].tick_params(axis='x', rotation=45)

Second subplot (Bottom categories)

sns.violinplot( data=df[df['category'].isin(categories_bottom)], x='category', y='num_subscribers', ax=axes[1] ) axes[1].set_ylim(0, 50000) axes[1].get_yaxis().set_major_formatter(plt.FuncFormatter(lambda x, _: f'{int(x):,}'))

axes[1].set_title('Subscribers vs Category (Bottom Categories)') axes[1].set_ylabel('Number of Subscribers') axes[1].set_xlabel('Category') axes[1].tick_params(axis='x', rotation=45)

plt.tight_layout() plt.show() sns.barplot(data=df,x='is_paid',y='num_subscribers',errorbar=None,estimator='mean') plt.ylabel('average num_subscribers')

multifeatured analysis

category-subscribers-is_paid

sns.barplot(data=df[df['is_paid']==False],x='category',y='num_subscribers',errorbar=None,estimator='mean')

plt.xlabel('category of unpaid courses') plt.ylabel('average subscribers') plt.title("filtering unpaid courses and then plotting category vs subscribers") plt.xticks(rotation=90) plt.show()

sns.barplot(data=df[df['is_paid']==True],x='category',y='num_subscribers',errorbar=None,estimator='mean')

plt.xlabel('category of paid courses') plt.ylabel('average subscribers') plt.title("filtering paid courses and then plotting category vs subscribers") plt.xticks(rotation=90) plt.show() grouped_data = ( df.groupby(['category', 'is_paid']) .agg(avg_subscribers=('num_subscribers', 'mean')) .reset_index() )

pivot_data = grouped_data.pivot(index='category', columns='is_paid', values='avg_subscribers').fillna(0)

pivot_data.plot(kind='bar', figsize=(12, 8), stacked=False, width=0.7)

plt.title('Number of Subscribers on avg by Category and is_paid per course', fontsize=14) plt.ylabel('Average Subscribers', fontsize=12) plt.xlabel('Category', fontsize=12) plt.legend(title='is_paid', fontsize=10) plt.xticks(rotation=45, ha='right')

plt.tight_layout() plt.show() grouped_data = ( df.groupby(['category', 'is_paid']) .agg(avg_subscribers=('num_subscribers', 'mean')) .reset_index() )

grouped_data['probability'] = grouped_data.groupby('category')['avg_subscribers'].transform(lambda x: x / x.sum())

pivot_data = grouped_data.pivot(columns='is_paid', index='category', values='probability')

pivot_data.plot(kind='bar', figsize=(12, 8), stacked=True, width=0.7)

plt.title('Number of Subscribers on avg by Category and is_paid per course', fontsize=14) plt.ylabel('probability of num_subscribers', fontsize=12) plt.xlabel('Category', fontsize=12) plt.legend(title='is_paid', fontsize=10) plt.xticks(rotation=45, ha='right')

plt.tight_layout() plt.show()

we plot avg subscriber per course vs category and instructional_level

levels=df.groupby('instructional_level') beginner_courses=levels.get_group('Beginner Level') all_level_courses=levels.get_group('All Levels') intermediate_courses=levels.get_group('Intermediate Level') expert_courses=levels.get_group('Expert Level') grouped_data = ( df.groupby(['category', 'instructional_level']) .agg(avg_subscribers=('num_subscribers', 'mean')) # Calculate mean .reset_index() )

pivot_data = grouped_data.pivot(columns='instructional_level', index='category', values='avg_subscribers')

pivot_data.plot(kind='bar', figsize=(12, 8), stacked=False, width=0.7)

plt.title('Number of Subscribers on avg by Category and Instructional Level per course', fontsize=14) plt.ylabel('Average Subscribers', fontsize=12) plt.xlabel('Category', fontsize=12) plt.legend(title='Instructional Level', fontsize=10) plt.xticks(rotation=45, ha='right')

plt.tight_layout() plt.show()

grouped_data = ( df.groupby(['category', 'instructional_level']) .agg(avg_subscribers=('num_subscribers', 'mean')) .reset_index() )

grouped_data['probability'] = grouped_data.groupby('category')['avg_subscribers'].transform(lambda x: x / x.sum())

pivot_data = grouped_data.pivot(columns='instructional_level', index='category', values='probability')

pivot_data.plot(kind='bar', figsize=(12, 8), stacked=False, width=0.7)

plt.title('Number of Subscribers(probabilities) by category and Instructional Level per course', fontsize=14) plt.ylabel('probability of no. of subscribers', fontsize=12)#ratio of no. of subscribers of that #level to the total no. of subscribers of the category across all levels plt.xlabel('Instructional Level', fontsize=12) plt.legend(title='category', fontsize=10) plt.xticks(rotation=45, ha='right')

Show the plot

plt.tight_layout() plt.show() grouped_data = ( df.groupby(['is_paid', 'instructional_level']) .agg(avg_subscribers=('num_subscribers', 'mean')) # Calculate mean .reset_index() )

pivot_data = grouped_data.pivot(index='instructional_level', columns='is_paid', values='avg_subscribers').fillna(0)

pivot_data.plot(kind='bar', figsize=(12, 8), stacked=False, width=0.7)

plt.title('Number of Subscribers on avg by is_paid and Instructional Level per course', fontsize=14) plt.ylabel('Average Subscribers', fontsize=12) plt.xlabel('Instructional Level', fontsize=12) plt.legend(title='is_paid', fontsize=10) plt.xticks(rotation=45, ha='right')

plt.tight_layout() plt.show()

Group by 'category' and 'instructional_level' and sum 'num_subscribers'

grouped_data = ( df.groupby(['is_paid', 'instructional_level']) .agg(avg_subscribers=('num_subscribers', 'mean')) # Calculate mean .reset_index() )

grouped_data['probability'] = grouped_data.groupby('instructional_level')['avg_subscribers'].transform(lambda x: x / x.sum())

pivot_data = grouped_data.pivot(index='instructional_level', columns='is_paid', values='probability').fillna(0)

pivot_data.plot(kind='bar', figsize=(12, 8), stacked=True, width=0.7)

plt.title('Number of Subscribers(probabilities) by is_paid and Instructional Level per course', fontsize=14) plt.ylabel('probability of num_subscribers', fontsize=12) plt.xlabel('Instructional Level', fontsize=12) plt.legend(title='is_paid', fontsize=10) plt.xticks(rotation=45, ha='right')

plt.tight_layout() plt.show()

Using Instructor_names and combine with other features to see which instructors have given hit courses

doing basic analysis instructor names

pd.options.display.float_format = '{:.2f}'.format df['num_subscribers'].describe() grby_category=df.groupby('category') photo_courses=grby_category.get_group('Photography & Video') finance_courses=grby_category.get_group('Finance & Accounting') fitness_courses=grby_category.get_group('Health & Fitness') development_courses=grby_category.get_group('Development') teaching_courses=grby_category.get_group('Teaching & Academics') PersDevelopment_courses=grby_category.get_group('Personal Development') design_courses=grby_category.get_group('Design') lifestyle_courses=grby_category.get_group('Lifestyle') marketing_courses=grby_category.get_group('Marketing') music_courses=grby_category.get_group('Music') officeProductivity_courses=grby_category.get_group('Office Productivity') business_courses=grby_category.get_group('Business') software_courses=grby_category.get_group('IT & Software')

analysing succesful music instructors

music_courses['num_subscribers'].describe() #filtering superhit music courses filt1=(music_courses['num_subscribers']>50000) & (music_courses['rating']>=4.5) df1=music_courses[filt1].reset_index(drop=True) df1 df1['instructor_names'] music_courses['rating'].describe()

filtering intermediate and expert level courses which are not that hit in num_subscribers but they have good rating

filt2=(music_courses['rating']==5) & (music_courses['num_subscribers']>500) & ((music_courses['instructional_level']=='Expert Level') | (music_courses['instructional_level']=='Intermediate Level')) df2=music_courses[filt2] df2 df2['instructor_names']

anaysing photography and video courses

photo_courses['rating'].describe() photo_courses['num_subscribers'].describe() #filtering superhit photography courses df3=photo_courses[(photo_courses['num_subscribers']>=100000) & (photo_courses['rating']>=4.5)] df3 pd.set_option('display.max_colwidth', None) print(df3['instructor_names']) pd.reset_option('display.max_colwidth')

filtering intermediate and expert level courses which are not that hit in num_subscribers but they have good rating

filt3=(photo_courses['rating']==5) & (photo_courses['num_subscribers']>500) & ((photo_courses['instructional_level']=='Expert Level') | (photo_courses['instructional_level']=='Intermediate Level')) photo_courses[filt3]

analyzing lifestyle courses

lifestyle_courses['num_subscribers'].describe()

check for instructors of superhit courses

lifestyle_courses[lifestyle_courses['num_subscribers']>60000]

filtering intermediate and expert level courses which are not that hit in num_subscribers but they have good rating

lifestyle_courses[(lifestyle_courses['rating']==5) & (lifestyle_courses['num_subscribers']>800) & ((lifestyle_courses['instructional_level']=='Expert Level') | (lifestyle_courses['instructional_level']=='Intermediate Level'))]

analysing fitness courses

fitness_courses['num_subscribers'].describe() #filtering superhit courses fitness_courses[fitness_courses['num_subscribers']>70000]

filtering intermediate and expert level courses which are not that hit in num_subscribers but they have good rating

fitness_courses[(fitness_courses['rating']==5) & (fitness_courses['num_subscribers']>800) & ((fitness_courses['instructional_level']=='Expert Level') | (fitness_courses['instructional_level']=='Intermediate Level'))]

analysing teaching courses

teaching_courses['num_subscribers'].describe() #filtering most hit teaching courses teaching_courses[teaching_courses['num_subscribers']>120000]

filtering intermediate and expert level courses which are not that hit in num_subscribers but they have good rating

teaching_courses[(teaching_courses['rating']==5) & (teaching_courses['num_subscribers']>2000) & ((teaching_courses['instructional_level']=='Expert Level') | (teaching_courses['instructional_level']=='Intermediate Level'))]

analysing personal development courses

PersDevelopment_courses['num_subscribers'].describe() #filtering most hit courses PersDevelopment_courses[PersDevelopment_courses['num_subscribers']>120000]

filtering intermediate and expert level courses which are not that hit in num_subscribers but they have excellent rating

PersDevelopment_courses[(PersDevelopment_courses['rating']==5) & (PersDevelopment_courses['num_subscribers']>1100) & ((PersDevelopment_courses['instructional_level']=='Expert Level') | (PersDevelopment_courses['instructional_level']=='Intermediate Level'))]

analysing finance courses

finance_courses['num_subscribers'].describe() #filtering most hit finance courses finance_courses[finance_courses['num_subscribers']>160000]

filtering intermediate and expert level courses which are not that hit in num_subscribers but they have excellent rating

finance_courses[(finance_courses['rating']==5) & (finance_courses['num_subscribers']>1030) & ((finance_courses['instructional_level']=='Expert Level') | (finance_courses['instructional_level']=='Intermediate Level'))]

analysing design courses

design_courses['num_subscribers'].describe() #filtering most hit design courses design_courses[design_courses['num_subscribers']>200000]

filtering intermediate and expert level courses which are not that hit in num_subscribers but they have excellent rating

design_courses[(design_courses['rating']==5) & (design_courses['num_subscribers']>900) & ((design_courses['instructional_level']=='Expert Level') | (design_courses['instructional_level']=='Intermediate Level'))]

analysing marketing courses

marketing_courses['num_subscribers'].describe() #filtering most hit marketing courses marketing_courses[marketing_courses['num_subscribers']>220000]

notice that highest subscriber is 8 lakh where 2nd highest is 3 lakh 40 thousand

filtering intermediate and expert level courses which are not that hit in num_subscribers but they have excellent rating

marketing_courses[(marketing_courses['rating']==5) & (marketing_courses['num_subscribers']>900) & ((marketing_courses['instructional_level']=='Expert Level') | (marketing_courses['instructional_level']=='Intermediate Level'))]

analysing office productivity courses

officeProductivity_courses['num_subscribers'].describe() #filtering most hit courses officeProductivity_courses[officeProductivity_courses['num_subscribers']>200000]

notice the course has highest subscriber as 15 lakh and 2nd highest 7 lakh 25 thousand

filtering intermediate and expert level courses which are not that hit in num_subscribers but they have excellent rating

officeProductivity_courses[(officeProductivity_courses['rating']>=4.9) & (officeProductivity_courses['num_subscribers']>700) & ((officeProductivity_courses['instructional_level']=='Expert Level') | (officeProductivity_courses['instructional_level']=='Intermediate Level'))]

analysing business courses

business_courses['num_subscribers'].describe() #filtering most hit business courses business_courses[business_courses['num_subscribers']>350000]

notice that all hit courses are 'All Level' courses

filtering intermediate and expert level courses which are not that hit in num_subscribers but they have excellent rating

business_courses[(business_courses['rating']==5) & (business_courses['num_subscribers']>900) & ((business_courses['instructional_level']=='Expert Level') | (business_courses['instructional_level']=='Intermediate Level'))]

analysing software courses

pd.options.display.float_format = '{:.2f}'.format software_courses['num_subscribers'].describe() #filtering most hit courses software_courses[software_courses['num_subscribers']>450000]

filtering intermediate and expert level courses which are not that hit in num_subscribers but they have excellent rating

software_courses[(software_courses['rating']==5) & (software_courses['num_subscribers']>900) & ((software_courses['instructional_level']=='Expert Level') | (software_courses['instructional_level']=='Intermediate Level'))]

analysing development courses

development_courses['num_subscribers'].describe() #filtering most hit development courses development_courses[development_courses['num_subscribers']>900000]

filtering intermediate and expert level courses which are not that hit in num_subscribers but they have excellent rating

development_courses[(development_courses['rating']>=4.9) & (development_courses['num_subscribers']>2000) & ((development_courses['instructional_level']=='Expert Level') | (development_courses['instructional_level']=='Intermediate Level'))]

Creating subcategories

creating subcategories for music courses pd.set_option('display.max.colwidth',None) pd.set_option('display.max.rows',3854) music_courses['title'] pd.set_option('display.max.colwidth',None) pd.set_option('display.max.rows',3854) music_courses.loc[:,['title','objectives']] musiccourses_new=music_courses.copy() subcategories = { 'music production steps and softwares/tools': ['reason','avid','cubase','akai','sibelius','serum','fl studio', 'garageband','cubase','maschine','personus','equalization', 'equalize','record','recording','compose','studio one','compress','compression', 'orchestration','sound design','ableton','logicpro','logic pro', 'logic pro x','audio mastering','audio production','garage band'], 'DJ & EDM': ['dj','edm','remix','mix','remixes','mixes','traktor'], 'music theory': ['electronic music','bass','harmony','abrsm'], 'singing and listening music':['vocal','sing','singing','breathing','write','songwriter','song writer', 'voice','throat','aural','hearing','ear'], 'guitar':['guitar'], 'piano':['piano'], 'Harmonium':['Harmonium','Harmonica'], 'music healing':['healing','therapy','relaxing','treat','treating'], 'indian music':['hindustani','indian','bansuri','kartal','mradanga','tabla','sitar','bharat','bhartiya'], 'ukulele':['ukulele'], 'other musical instruments':['violin','flute','trumpet','drum','drummers','saxophone','sax','cellophane' 'cello','cellophone','mandolin','banjo','clarinet','bouzouki','oscarina', 'ocarina']

} def assign_subcategory(row): text = ' '.join([ str(row['title']), str(row['objectives']), ]).lower()

for subcategory, keywords in subcategories.items():
    if any(keyword.lower() in text for keyword in keywords):
        return subcategory
return 'Other'

Apply the function to the dataset

musiccourses_new['subcategory'] = musiccourses_new.apply(assign_subcategory, axis=1) musiccourses_new.head() musiccourses_new['subcategory'].value_counts() sns.barplot(data=musiccourses_new,x='subcategory',y='num_subscribers',errorbar=None,estimator='mean') plt.ylabel('average num_subscribers') plt.xticks(rotation=90) plt.show() grouped_data = ( musiccourses_new.groupby(['subcategory', 'is_paid']) .agg(avg_subscribers=('num_subscribers', 'mean')) .reset_index() )

pivot_data = grouped_data.pivot(index='subcategory', columns='is_paid', values='avg_subscribers')

pivot_data.plot(kind='bar', figsize=(12, 8), stacked=False, width=0.7)

plt.title('Number of Subscribers on avg by subategory and is_paid per course', fontsize=14) plt.ylabel('Average Subscribers', fontsize=12) plt.xlabel('subcategory', fontsize=12) plt.legend(title='is_paid', fontsize=10) plt.xticks(rotation=90, ha='right')

plt.tight_layout() plt.show() grouped_data = ( musiccourses_new.groupby(['subcategory', 'instructional_level']) .agg(avg_subscribers=('num_subscribers', 'mean')) .reset_index() )

pivot_data = grouped_data.pivot(index='subcategory', columns='instructional_level', values='avg_subscribers')

pivot_data.plot(kind='bar', figsize=(12, 8), stacked=False, width=0.7)

plt.ylabel('Average Subscribers', fontsize=12) plt.xlabel('subcategory', fontsize=12) plt.legend(title='instructional_level', fontsize=10) plt.xticks(rotation=90, ha='right')

plt.tight_layout() plt.show()

creating subcategory for development courses pd.set_option('display.max.colwidth',9945) pd.set_option('display.max.rows',9945) development_courses[['title','objectives']] pd.reset_option('display.max.colwidth') developmentcourses_new=development_courses.copy() subcategory_keywords = { "Web Development & Database": [ {"html"}, {"css"}, {"javascript"}, {"react"}, {"frontend"}, {"backend"},{'scala','web'},{'java','web'}, {"nextjs"}, {"expressjs"}, {"tailwind"}, {"angular"}, {"nodejs"}, {"web development"}, {"spring boot"}, {"mongodb"}, {"mongo db"}, {"sql"}, {"postgres"}, {"wordpress"}, {"full stack"}, {"mern stack"}, {"php"}, {"database"}, {"rust", "web"} ], "AI/ML_Data": [ {"machine learning"}, {"ml"}, {"ai"}, {"deep learning"}, {"neural network"},{'backpropagation'}, {"data science"}, {"spark"}, {"pytorch"}, {"numpy"}, {"pandas"}, {"seaborn"},{'artificial intelligence'}, {"matplotlib"}, {"tensorflow"}, {"tableau"}, {"power bi"}, {"chatgpt"},{'scala','data'}, {"openai"}, {"gen ai"}, {"llm"}, {"big data"}, {"eda"}, {"data analysis"}, {"apache"}, {"nlp"}, {"computer vision"}, {"sklearn"}, {"scikit learn"}, {"statistics"}, {"calculus"}, {"pyspark"} ], "Blockchain": [{"blockchain"}, {"etherium"}], "Android Development": [ {"android"}, {"ios"}, {"kotlin"}, {"mobile app"}, {"flutter"}, {"dart"}, {"appium"}, {"xamarin"} ], "DevOps": [ {"devops"}, {"docker"}, {"kubernetes"}, {"ci/cd"}, {"continuous integration"}, {"jenkins"}, {"gitlab"} ], "Web Scraping": [{"selenium"}, {"beautiful soup"}], "Cloud": [{"aws"}, {"azure"}], "Cyber Security": [{"bug"}, {"bug testing"}], "Python_Web": [{"python"},{'flask'},{'django'},{'fastapi'}], "data_structure & other_prog_languages ": [{"data structures",'algorithm'},{"data structures",'algorithms'}, {"object oriented"}, {"oops"},{'c'},{'c++'},{'c#'}], "Version Control": [{"git"}, {"github"}], "Game Development": [{"unity"}, {"game"}, {"unreal game engine"},{'unreal engine'}] }

def assign_subcategory(row, subcategory_keywords): text = f"{row['title']} {row['objectives']} ".lower() for subcategory, keyword_sets in subcategory_keywords.items(): for keywords in keyword_sets: if all(keyword in text for keyword in keywords): return subcategory return "Other"

developmentcourses_new["subcategory"] = developmentcourses_new.apply(assign_subcategory, axis=1, args=(subcategory_keywords,)) developmentcourses_new.head() developmentcourses_new['subcategory'].value_counts() pd.set_option('display.max.colwidth',None) developmentcourses_new[developmentcourses_new['subcategory']=='Other'][['title','objectives']] sns.barplot(data=developmentcourses_new,x='subcategory',y='num_subscribers',estimator='mean',errorbar=None) plt.xticks(rotation=90) plt.ylabel('subscribers on average') plt.show() grouped_data = ( developmentcourses_new.groupby(['subcategory', 'instructional_level']) .agg(avg_subscribers=('num_subscribers', 'mean')) .reset_index() )

pivot_data = grouped_data.pivot(columns='instructional_level', index='subcategory', values='avg_subscribers')

pivot_data.plot(kind='bar', figsize=(12, 8), stacked=False, width=0.7)

plt.title('Number of Subscribers on avg by subcategory and instructional_level per course', fontsize=14) plt.ylabel('Average Subscribers', fontsize=12) plt.xlabel('subcategory', fontsize=12) plt.legend(title='instructional_level', fontsize=10) plt.xticks(rotation=90, ha='right')

plt.tight_layout() plt.show() grouped_data = ( developmentcourses_new.groupby(['subcategory', 'instructional_level']) .agg(avg_subscribers=('num_subscribers', 'mean')) .reset_index() )

grouped_data['probability'] = grouped_data.groupby('subcategory')['avg_subscribers'].transform(lambda x: x / x.sum())

pivot_data = grouped_data.pivot(columns='instructional_level', index='subcategory', values='probability') pivot_data.plot(kind='bar', figsize=(12, 8), stacked=False, width=0.7)

plt.title('Number of Subscribers on avg by subcategory and instructional_level per course', fontsize=14) plt.ylabel('probability of Subscribers', fontsize=12) plt.xlabel('subcategory', fontsize=12) plt.legend(title='instructional_level', fontsize=10) plt.xticks(rotation=90, ha='right')

plt.tight_layout() plt.show()