🚀 API Reference

The core of this project is a serverless API endpoint that allows you to get a view count prediction based on video metadata.

Predict Endpoint

• HTTP Method: POST
• Endpoint URL: https://3bd5z76vu0.execute-api.us-east-1.amazonaws.com/v1/predict
• Description: Sends video feature data to the model and returns a predicted view count.

---

Request Body (JSON)

The POST request must include a JSON body with the following structure and data types.

Parameter	Type	Description	Example
like_count	integer	The total number of likes on the video.	5000
comment_count	integer	The total number of comments on the video.	200
duration_seconds	integer	The length of the video in seconds.	300
tag_count	integer	The number of tags associated with the video.	15
category_id	integer	The numerical ID for the YouTube category (e.g., 24 is Entertainment).	24
publish_hour	integer	The hour of the day the video was published (0-23).	14
publish_day_of_week	integer	The day of the week (0=Monday, 6=Sunday).	3
channel_title	string	The name of the YouTube channel.	"Some Channel"

Example Request Body

{
  "like_count": 5000,
  "comment_count": 200,
  "duration_seconds": 300,
  "tag_count": 15,
  "category_id": 24,
  "publish_hour": 14,
  "publish_day_of_week": 3,
  "channel_title": "Some Channel"
}

Responses

Successful Response (200 OK)

If the request is successful, the API will return a JSON object with the predicted view count.

{
  "predicted_views": 1234567.89
}

Error Response (400 Bad Request)

If the input data is missing or invalid, the API may return an error.

{
  "error": "Invalid input provided. Please check your JSON payload."
}