Instagram Engagement Analysis¶
Executive Summary¶
This project analyzes Instagram post engagement to identify content performance patterns.
Key Findings: • Photo posts slightly outperform video and carousel posts. • Moderate hashtag usage (2–3 hashtags) maintains stable engagement. • Emoji usage is consistent across all posts. • Posting time variation was limited in dataset.
Recommendations:
- Prioritize high-quality photo content.
- Maintain balanced content mix.
- Use 2–5 targeted hashtags.
- Continue emoji usage.
- Implement structured weekly content calendar.
In [2]:
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
In [4]:
users = pd.read_csv("C:\\Users\\DeLL\\OneDrive\\Desktop\\Alfido Tech Internship\\Instagram\\users.csv")
photos = pd.read_csv("C:\\Users\\DeLL\\OneDrive\\Desktop\\Alfido Tech Internship\\Instagram\\photos.csv")
likes = pd.read_csv("C:\\Users\\DeLL\\OneDrive\\Desktop\\Alfido Tech Internship\\Instagram\\likes.csv")
comments = pd.read_csv("C:\\Users\\DeLL\\OneDrive\\Desktop\\Alfido Tech Internship\\Instagram\\comments.csv")
tags = pd.read_csv("C:\\Users\\DeLL\\OneDrive\\Desktop\\Alfido Tech Internship\\Instagram\\tags.csv")
photo_tags = pd.read_csv("C:\\Users\\DeLL\\OneDrive\\Desktop\\Alfido Tech Internship\\Instagram\\photo_tags.csv")
follows = pd.read_csv("C:\\Users\\DeLL\\OneDrive\\Desktop\\Alfido Tech Internship\\Instagram\\follows.csv")
In [6]:
users.head()
photos.head()
likes.head()
comments.head()
Out[6]:
| id | comment | User id | Photo id | created Timestamp | posted date | emoji used | Hashtags used count | |
|---|---|---|---|---|---|---|---|---|
| 0 | 1 | unde at dolorem | 2 | 1 | 13-04-2023 08:04 | April 14 | yes | 1 |
| 1 | 2 | quae ea ducimus | 3 | 1 | 13-04-2023 08:04 | April 14 | no | 2 |
| 2 | 3 | alias a voluptatum | 5 | 1 | 13-04-2023 08:04 | April 14 | no | 4 |
| 3 | 4 | facere suscipit sunt | 14 | 1 | 13-04-2023 08:04 | April 14 | yes | 2 |
| 4 | 5 | totam eligendi quaerat | 17 | 1 | 13-04-2023 08:04 | April 14 | yes | 1 |
In [8]:
users.shape
photos.shape
likes.shape
comments.shape
Out[8]:
(7488, 8)
In [10]:
print("Users:", users.shape)
print("Photos:", photos.shape)
print("Likes:", likes.shape)
print("Comments:", comments.shape)
print("Tags:", tags.shape)
print("Photo_Tags:", photo_tags.shape)
print("Follows:", follows.shape)
Users: (100, 6) Photos: (257, 6) Likes: (8782, 5) Comments: (7488, 8) Tags: (21, 4) Photo_Tags: (501, 3) Follows: (7623, 5)
In [14]:
follows.columns
Out[14]:
Index(['follower', 'followee ', 'created time', 'is follower active',
'followee Acc status'],
dtype='object')
In [18]:
follows.columns = follows.columns.str.strip()
likes.columns = likes.columns.str.strip()
comments.columns = comments.columns.str.strip()
photos.columns = photos.columns.str.strip()
users.columns = users.columns.str.strip()
photo_tags.columns = photo_tags.columns.str.strip()
tags.columns = tags.columns.str.strip()
In [20]:
follows.columns
Out[20]:
Index(['follower', 'followee', 'created time', 'is follower active',
'followee Acc status'],
dtype='object')
In [22]:
followers_count = follows.groupby('followee').size().reset_index(name='followers')
followers_count.head()
Out[22]:
| followee | followers | |
|---|---|---|
| 0 | 1 | 77 |
| 1 | 2 | 76 |
| 2 | 3 | 76 |
| 3 | 4 | 76 |
| 4 | 5 | 76 |
In [26]:
likes.columns
Out[26]:
Index(['user', 'photo', 'created time', 'following or not', 'like type'], dtype='object')
In [30]:
likes_count = likes.groupby('photo').size().reset_index(name='likes_count')
likes_count.head()
Out[30]:
| photo | likes_count | |
|---|---|---|
| 0 | 1 | 25 |
| 1 | 2 | 36 |
| 2 | 3 | 38 |
| 3 | 4 | 38 |
| 4 | 5 | 31 |
In [32]:
comments.columns
Out[32]:
Index(['id', 'comment', 'User id', 'Photo id', 'created Timestamp',
'posted date', 'emoji used', 'Hashtags used count'],
dtype='object')
In [34]:
comments_count = comments.groupby('Photo id').size().reset_index(name='comments_count')
comments_count.head()
Out[34]:
| Photo id | comments_count | |
|---|---|---|
| 0 | 1 | 25 |
| 1 | 2 | 31 |
| 2 | 3 | 27 |
| 3 | 4 | 32 |
| 4 | 5 | 27 |
In [36]:
photos.columns
Out[36]:
Index(['id', 'image link', 'user ID', 'created dat', 'Insta filter used',
'photo type'],
dtype='object')
In [38]:
comments_count = comments.groupby('Photo id').size().reset_index(name='comments_count')
In [40]:
photos_merged = photos.merge(
likes_count,
left_on='id',
right_on='photo',
how='left'
)
photos_merged = photos_merged.merge(
comments_count,
left_on='id',
right_on='Photo id',
how='left'
)
photos_merged = photos_merged.merge(
followers_count,
left_on='user ID',
right_on='followee',
how='left'
)
In [42]:
photos_merged[['likes_count','comments_count','followers']] = \
photos_merged[['likes_count','comments_count','followers']].fillna(0)
In [44]:
photos_merged['engagement_rate'] = (
photos_merged['likes_count'] +
photos_merged['comments_count']
) / photos_merged['followers']
photos_merged['engagement_rate'] = \
photos_merged['engagement_rate'].replace([np.inf, -np.inf], 0)
In [46]:
photos_merged[['likes_count','comments_count','followers','engagement_rate']].describe()
Out[46]:
| likes_count | comments_count | followers | engagement_rate | |
|---|---|---|---|---|
| count | 257.000000 | 257.000000 | 257.000000 | 257.000000 |
| mean | 34.171206 | 29.136187 | 76.307393 | 0.829693 |
| std | 3.605804 | 3.452939 | 0.462314 | 0.063600 |
| min | 25.000000 | 21.000000 | 76.000000 | 0.649351 |
| 25% | 31.000000 | 27.000000 | 76.000000 | 0.789474 |
| 50% | 34.000000 | 29.000000 | 76.000000 | 0.828947 |
| 75% | 37.000000 | 31.000000 | 77.000000 | 0.868421 |
| max | 48.000000 | 39.000000 | 77.000000 | 1.039474 |
In [52]:
photos_merged['created dat'] = pd.to_datetime(
photos_merged['created dat'],
format='%d-%m-%Y %H:%M'
)
In [54]:
photos_merged['hour'] = photos_merged['created dat'].dt.hour
photos_merged['day_name'] = photos_merged['created dat'].dt.day_name()
In [56]:
hourly_engagement = photos_merged.groupby('hour')['engagement_rate'].mean().reset_index()
hourly_engagement.sort_values(by='engagement_rate', ascending=False).head()
Out[56]:
| hour | engagement_rate | |
|---|---|---|
| 0 | 8 | 0.829693 |
In [60]:
plt.figure(figsize=(10,5))
sns.lineplot(data=hourly_engagement, x='hour', y='engagement_rate')
plt.title("Average Engagement by Posting Hour")
plt.show()
In [62]:
photos_merged['hour'].value_counts()
Out[62]:
hour 8 257 Name: count, dtype: int64
In [64]:
photo_type_engagement = photos_merged.groupby('photo type')['engagement_rate'].mean().reset_index()
photo_type_engagement.sort_values(by='engagement_rate', ascending=False)
Out[64]:
| photo type | engagement_rate | |
|---|---|---|
| 1 | photo | 0.833777 |
| 2 | video | 0.826890 |
| 0 | carousel | 0.819945 |
In [66]:
plt.figure(figsize=(8,5))
sns.barplot(data=photo_type_engagement, x='photo type', y='engagement_rate')
plt.xticks(rotation=45)
plt.title("Engagement by Photo Type")
plt.show()
In [68]:
comments[['Photo id','Hashtags used count']].head()
Out[68]:
| Photo id | Hashtags used count | |
|---|---|---|
| 0 | 1 | 1 |
| 1 | 1 | 2 |
| 2 | 1 | 4 |
| 3 | 1 | 2 |
| 4 | 1 | 1 |
In [70]:
hashtag_count = comments.groupby('Photo id')['Hashtags used count'].mean().reset_index()
photos_merged = photos_merged.merge(
hashtag_count,
left_on='id',
right_on='Photo id',
how='left'
)
photos_merged['Hashtags used count'] = \
photos_merged['Hashtags used count'].fillna(0)
In [72]:
hashtag_engagement = photos_merged.groupby('Hashtags used count')['engagement_rate'].mean().reset_index()
hashtag_engagement.sort_values(by='engagement_rate', ascending=False)
Out[72]:
| Hashtags used count | engagement_rate | |
|---|---|---|
| 20 | 2.358974 | 1.039474 |
| 12 | 2.323529 | 0.986842 |
| 36 | 2.410256 | 0.935065 |
| 33 | 2.405405 | 0.921053 |
| 46 | 2.441176 | 0.916667 |
| ... | ... | ... |
| 57 | 2.480000 | 0.751623 |
| 56 | 2.478261 | 0.743421 |
| 30 | 2.391304 | 0.714286 |
| 4 | 2.260870 | 0.684211 |
| 26 | 2.380952 | 0.662338 |
78 rows × 2 columns
In [74]:
plt.figure(figsize=(10,5))
sns.lineplot(data=hashtag_engagement, x='Hashtags used count', y='engagement_rate')
plt.title("Engagement vs Hashtag Count")
plt.show()
In [76]:
photos_merged['hashtag_bucket'] = \
photos_merged['Hashtags used count'].round()
In [78]:
hashtag_engagement_bucket = \
photos_merged.groupby('hashtag_bucket')['engagement_rate'].mean().reset_index()
hashtag_engagement_bucket.sort_values(by='engagement_rate', ascending=False)
Out[78]:
| hashtag_bucket | engagement_rate | |
|---|---|---|
| 0 | 2.0 | 0.830606 |
| 1 | 3.0 | 0.823899 |
In [80]:
plt.figure(figsize=(8,5))
sns.barplot(data=hashtag_engagement_bucket,
x='hashtag_bucket',
y='engagement_rate')
plt.title("Engagement by Hashtag Count")
plt.show()
In [82]:
emoji_usage = comments.groupby('Photo id')['emoji used'].apply(
lambda x: (x == 'yes').sum()
).reset_index(name='emoji_count')
In [84]:
photos_merged = photos_merged.merge(
emoji_usage,
left_on='id',
right_on='Photo id',
how='left'
)
photos_merged['emoji_count'] = photos_merged['emoji_count'].fillna(0)
In [86]:
photos_merged['emoji_bucket'] = photos_merged['emoji_count'].apply(
lambda x: 'With Emoji' if x > 0 else 'No Emoji'
)
emoji_engagement = photos_merged.groupby('emoji_bucket')['engagement_rate'].mean().reset_index()
emoji_engagement
Out[86]:
| emoji_bucket | engagement_rate | |
|---|---|---|
| 0 | With Emoji | 0.829693 |
In [88]:
plt.figure(figsize=(6,4))
sns.barplot(data=emoji_engagement,
x='emoji_bucket',
y='engagement_rate')
plt.title("Engagement: Emoji vs No Emoji")
plt.show()
In [90]:
photos_merged['emoji_bucket'].value_counts()
Out[90]:
emoji_bucket With Emoji 257 Name: count, dtype: int64
8. Strategy Recommendations¶
Based on the engagement analysis of 257 posts, the following strategic insights are recommended for Alfido Tech:
- Prioritize high-quality photo posts, as they show slightly higher engagement compared to videos and carousels.
- Maintain a balanced content mix, including educational videos and carousel guides.
- Use 2–5 relevant and targeted hashtags to maintain stable engagement without overloading posts.
- Continue emoji usage to maintain conversational tone and relatability.
- Conduct A/B testing on posting time (8 AM vs evening slots) to identify optimal engagement windows.
These strategies aim to improve content consistency, audience interaction, and brand visibility.
9. Conclusion¶
This analysis examined content type performance, hashtag usage, emoji impact, and posting schedule patterns.
While the dataset showed limited variation in posting time and emoji usage, key engagement trends were identified. Photo-based content slightly outperformed other formats, and moderate hashtag usage maintained stable engagement levels.
Implementing a structured content calendar and controlled posting experiments can further enhance Instagram growth and audience engagement for Alfido Tech.