Build a customer classification model

Good day! I will complete this task quickly, and most importantly, with quality in 1 day!!!! Feel free to contact me!!!!

Show original

Translate

Good evening! I did customer segmentation in Python — features in the style of RFM (recency, frequency, monetary, activity) + clustering and profiling of each group to create live segments instead of nameless numbers. I will extract data from Mongo/SQL, clean the raw records, and build the pipeline. First of all, what is approximately in those 20k records and what you need the segments for — I will select features based on that. I will complete it in 3 days.

Show original

Translate

Hello.

I can create the first working version of an ML pipeline in Python for customer segmentation based on behavioral characteristics.

A key clarification at the start: are there already prepared group labels? If so, we will perform supervised classification with a train/test split and quality metrics. If there are no labels, we will build segmentation through clustering, selecting the number of groups and describing the logic of each segment.

In the work process:
- data extraction from MongoDB / SQL or from exports
- cleaning raw records, handling missing values, anomalies
… - feature engineering: activity, frequency, recency, volumes, RFM-like features
- baseline model using scikit-learn
- quality assessment, saving the model and labeled dataset
- a brief report to clarify what each customer group means

Before starting the work, I need to clarify:
- what is the data structure in MongoDB / SQL
- are there predefined customer classes
- what result is needed: file, model, script, or writing groups back to the database
- how these groups will be used further

We can discuss the details in private messages.

Show original

Translate

Hello! I have worked on similar tasks in Python. Please clarify if these are unknown groups; do we need to find if there are existing group labels? This will determine the approach.

Show original

Translate

Good day!

The task is clear — building behavioral segmentation of clients based on raw data.

What I will do:

Connecting to MongoDB/SQL, extracting and auditing 20k records
Data cleaning: anomalies, missing values, normalization
Feature Engineering: from behavioral logs — activity metrics, frequency, volumes, RFM features
… Model: clustering (K-Means + DBSCAN) to identify behavioral groups, selecting the optimal number of clusters (elbow method + silhouette score)
Result: labeled dataset, saved model, report describing each client group and recommendations
Stack: Python, pandas, scikit-learn, pymongo

Clarify: are the data already cleaned or completely raw? Is there a target label (supervised) or do I need to find groups independently (unsupervised)?

Show original

Translate

Hello! The task of building a behavioral model of clients is completely clear to me. I am studying in the field of "Artificial Intelligence Systems," so developing the architecture of ML models and Feature Engineering is my professional stack.

Here’s how I propose to implement the project in Python:
Data Prep: We will extract raw data from your database (SQL/Mongo), perform cleaning from anomalies, and fill in missing values.
Feature Engineering: Based on behavioral logs, we will generate a dense feature matrix (activity metrics, frequency, and volumes of interaction).
Modeling: We will build and train the optimal model (classification/clustering using scikit-learn), tune hyperparameters, and derive metrics for evaluating the quality of the model.
I would be happy to discuss the specifics of the raw data in private messages!

Show original

Translate

Hello!

Based on your 20,000 raw records from Mongo/SQL, I will develop an optimized pipeline for data cleaning and generating behavioral features in Python. I will implement an accurate model that segments customers into groups and updates the status of each user in the database as needed. You will receive a ready-to-use documented script that automatically loads raw data, forms a feature matrix, and predicts classes. Could you let me know if you already have labeled classes for training the model, or is the task to find hidden segments through clustering? Please write to me privately — we will discuss the database structure and get started.

Show original

Translate