I have a dataframe with shop coordinates and want to divide them into clusters based on the day a vendor should visit that shop. For example, let's say that there are 180 shops that a vendor should visit. He should visit 30-34 shops from Monday to Friday and, on Saturday, he should visit 60% of the other days. How can I do that? Using kmeans-constrained I can only divide them into equally sized clusters. Maybe I need to use some kind of solver or move points between clusters to reach the numbers I want but I don't know how to do that. Here is the code that divides them equally: # Loop through the vendor clusters for vendor in df["vendor"].unique(): # Number of shops for each vendor n_shops = df.loc[df["vendor"] == vendor]["cod_shop"].count() # Index idx = df.loc[df["vendor"] == vendor].index # Number of clusters for days of the week num_clusters = 5 # Monday to Friday # Average size for the cluster avg_size = n_shops / (num_clusters + 0.6) # Define limits min_shops = round(avg_size - n_shops * pct, 0) max_shops = math.ceil(avg_size + n_shops * pct) # Model kmeans = KMeansConstrained(n_clusters=num_clusters, size_min=min_shops, size_max=max_shops, random_state=42) labels = kmeans.fit_predict(df.loc[df["vendor"] == vendor][["latitude", "longitude"]]) # Add labels to the dataframe df.loc[idx, "visit_day"] = labels Continue reading...
