Let say I have this dataset 
kor_cat is the categories of the product like(noodle,cookie....)
and the kor_qty is the number of product selled
I want to know the weather has a relation with the store's sale record
but I want to know which machine learning technique is fit here
and I have the categori of the product and number of selled of that product.
How should I solve this problem with ML.
Apparently, you have a target variable (y) and a set of input variables (input variables (X). This seems like a supervised problem. A regression analysis can be a solution to implement.
Feature engineering needs to be completed before running a machine learning model. Such as text field columns can be encoded with Label Encoder and One Hot Encoder.
Irrelevant input columns for analysis can be dropped. If the city variable is same all the way, meaning the all of the data is from a single city. Also N/A and null values need to be cleaned.
In some models, normalization of the data can produce better results.
As a Machine Learning model, starting from Linear Regression would be feasible. Later Random Forest Regressor, Lasso, Ridge Regression can be examined.
If you are interested in knowing what is the relation between two variables with ML, you should treat it as a supervised learning problem.
Use the store´s sales record as a target. Then fit any ML algorithm and check the feature importance of it.
If you have categorical features encode them.
Whilst at the outset it might look like a supervised learning problem one should look at the underlying business scenario as well here. Based on my personal experience I feel there could be categories of items which sell more during rainy days, some during sunny and others for cold/windy/snowy etc. My intuition is to understand these groups first either through clusters or even by looking at associate rule mining. Once you understand these clusters which hopefully would help separate the products which sell more for a particular weather type.
