是否有可能从数据集中获得两种不同类型的结果

Question

please be easy on me, I am new to ML. I am sure somebody will request to close this as subjective but I cannot find my specific answer and don't know how else to ask.

If I have a shop, with three areas of the shop. I have sensors to detect when people come in or out of each area. This happens every 15 seconds. So, in my db, I have a count of the occupancy, per room, every 15 seconds.

Using this data, I want to predict the occupancy, per room, in the future but also, if somebody comes in the door, predict most likely room they will go to.

Is it possible to predict future occupancy per room and also probability of where people will go when the walk in using a dataset that simply lists the rooms and the occupancy of each room every 15 seconds? Is this a regression model?

Thanks!

Mike

Answer 1

Predicting the most likely room, which they would walk in. :

This falls under the classification problem. The output falls under a set of categories, in this case it is different rooms.

Predicting the Occpancy of each room : As mentioned by @poorna is a regression problem.

Two ways you can look at this problem,

1. Multi- target regression problem with occupancy of each room as one target and past occupancies of all rooms as input.

2. Independent forecast problem for each room with past occupancies of corresponding room as input.

For learning the basics of machine learning, you can go through this link

Answer 2

I can see this is a forecasting (A type of Regression) problem.

This requires a set of features useful for forecasting your occupancy per room which can be

1. room number
2. time period as hours
3. time period as day of the week
4. Lagging

Try fitting any of the time series forecasting models like mentioned here with the above mentioned as features and occupancy count as target variable.

Basing on the count of occupancy per room you can find the most probable room by applying simple probability.