Hololive Production Livestream Activity Statistics

Made with Plotly and deployed in Streamlit.

Data was initialized using separate scripts published in another repo, HPLAS-Initialize.

Requirements

• Python
• Holodex API key (HOLODEX_API_KEY)
• YouTube Data API key (YT_DATA_API_KEY)

Setup

Create virtual environment:

python -m venv venv
source venv/bin/activate

Install libraries:

pip install -r requirements.txt

Run app locally:

streamlit run Main.py

JSON Schema

{
    "<name>": {
        "missing": int,
        "missing_length": int,
        "topics": {
            "<topic_name>": [int, int]
        },
        "details": [
            {
                "id": string,
                "title": string,
                "duration": int,
                "date": string
            }
        ]
    }
}

• name - Name of the channel. The same name used as the index in the CSV files.
  • missing - Amount of unarchived livestreams.
  • missing_length - Total duration of unarchived livestreams in seconds.
  • topics - The topic of the livestream provided by Holodex API.
    • topic_name - Name of the topic. Represented by an array of two integers. The 1st integer and 2nd integer is the amount of livestreams with the same topic and the total duration (in seoncds) of said livestreams respectively.
  • details - An array of JSON's for details of each livestream.
    • id - YouTube ID of the livestream.
    • title - Title of the livestream.
    • duration - Duration of the livestream.
    • date - The date and time the livestream started in ISO 8601 (UTC+0).

Update Process

Update is done manually by running scripts/update.py, preferrably every 2 weeks, max being 3 weeks. This is because the Holodex API only queries up to the latest 50 videos of a channel, including non-livestreams.

The script doesn't include the Hololive channel because it is incompatible to apply the 360 second duration filter (which mostly works on the members) since most of its videos are longer than that duration and aren't livestreams. When updating it, scripts/update_holo_ch.py is used (more documentation in the script). The channel is updated less frequently since it seldom livestreams.

Adding New Members

Think of a one-word name that will identify each member which is represented by name.

Add a row in data/details.csv most of the values are self-explanatory, but here are the explanations for those that are not:

• most, least, zero - The HEX colors for the heatmap that represent the amount of livestreams in a day of the week. Get the main color of the member from https://hololive.hololivepro.com/en/talents based on the darker color in their background, that will be the most color. Go to https://www.color-hex.com/ and enter the color. The 3rd from the right in the Tints section will be the least color. The complementary color will be the zero color. Not all heatmaps will have good contrast based on their main color so exercise discretion.

• active - 0 for not active member (graduated/terminated). 1 for active.

<name>,full_name,ch_name,ch_id,debut_date,most,least,zero,active,branch,gen_name

Add a row in data/data.csv:

<name>,,,,,,,,,,,,,,

Add in json/livestream_details.json:

"<name>":{"missing":0,"missing_length":0,"topics":{},"details":[]},

Add a row in data/generation_colors_name.csv. EN and ID have the same color as a branch. For new JP gens, pick the main color from any of the members in the gen that's not too similar to the current colors.

Put these templates after the latest gen of their branch so they look organized.