This table comes from Chapter 3 in Tom M. Mitchell. Machine Learning (free)

enter image description here

Is there a bigger dataset (discrete-valued) like this, to train a learner in making the PlayTennis determination?

The dataset should include:

  • Outlook, as a concise weather forecast like sunny/overcast/rain/snow
  • Temperature, generalized like hot/mild/cool/cold
  • Humidity, generalized like high/normal/low
  • Wind, generalized like weak/strong
  • PlayTennis - the determination whether to play tennis or not, based on the other variables
  • Are you just looking for the weather data, or do you also need the PlayTennis determination? – csk Dec 5 '19 at 17:37
  • @csk Thanks for your reminder. I've updated the post. PlayTennis determination – JJJohn Dec 6 '19 at 5:16
  • Thanks for the clarification. With the original question there was a chance people would just suggest datasets with only weather data in them, which are pretty easy to find, but your question is more specialized than that. I substantially edited your question to make it easier for people to help. If you disagree with any of my changes you can reject or improve on them by clicking the edit link under your question. – csk Dec 6 '19 at 16:24

My understanding is that PlayTennis itself is limited to 14 examples (see this version of the data on Kaggle), so if you wanted more you'd have to generate them yourself.

If you're not bound to this specific dataset, there exist many alternative datasets with categorical features used in discrete classification tasks. See that UCI Machine Learning repository. Try filtering to "Categorical" and/or "Mixed" attribute types and "Classification" for the default task. Some potential candidates datasets for your task (with mostly categorical features):

Note that it would be up to you to define the model in any of these cases (e.g. the features inputs and target outputs).

  • Thanks for your answer. It is very helpful. Does "discrete classification tasks" refer to any regular/general classification tasks, or a special classification tasks that runs on discrete features/attributes? of cause, the output/class label would be alway discrete. – JJJohn Dec 7 '19 at 8:40
  • I used the phrase here to indicate that in the case of the PlayTennis dataset, the classification task uses both discrete features and labels. While I'm sure this is highly dependent on the community you're in, my understanding has been that "discrete classification tasks" is refers to a problem where a model and data are used to predict a discrete set of outcomes. – sboysel Dec 7 '19 at 9:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.