|
import pandas as pd |
|
import streamlit as st |
|
|
|
st.title('Udemy Courses Analysis') |
|
|
|
courses_info = pd.read_csv('udemy_courses_info.csv') |
|
courses_year = pd.read_csv('udemy_courses_year.csv') |
|
|
|
couses_df = pd.merge(courses_info, courses_year, on='course_id') |
|
|
|
options = st.multiselect( |
|
"Please select subjects", |
|
couses_df.subject.unique(), |
|
couses_df.subject.unique()) |
|
|
|
couses_df = couses_df[couses_df['subject'].isin(options)] |
|
|
|
|
|
grouped = couses_df.groupby(['subject', 'course_title'])['num_subscribers'].sum() |
|
|
|
|
|
grouped = grouped.reset_index() |
|
|
|
|
|
grouped = grouped.sort_values(['subject', 'num_subscribers'], ascending=[True, False]) |
|
|
|
|
|
top_courses = grouped.groupby('subject').head(5) |
|
|
|
st.dataframe(top_courses) |