How to use a custom .csv dataset in TensorFlow Recommenders library?

user_6383522 0
Viewed 489

I'm new to tensorflow. I want to train a recommendation model on my dataset using the TensorFlow Recommenders library and the simple code provided at:

https://github.com/tensorflow/recommenders

I want to know how can I use (load and feed to the model) my custom .csv file in the following format instead of loading the built-in Movielens dataset?

user_id item_id rating
2 8 3
5 12 4
6 4 2
... ... ...

My Tensorflow versions:

tensorflow==2.4.0

tensorflow-datasets==4.2.0

tensorflow-recommenders==0.4.0

recommendation-system recommendation-engine tensorflow-datasets tensorflow python
1 Answers
import pandas as pd

DATA_URL = "D:/ratings.csv"
df = pd.read_csv(DATA_URL)
ratings = tf.data.Dataset.from_tensor_slices(dict(df)).map(lambda x: {
    "user_id": str(x["user_id"]),
    "item_id": str(x["item_id"]),
    "rating": float(x["rating"])
})
user_12224130
Related