I will forecast sales and demand using python time series
I turn data into decisions, churn, forecasting, customer analytics
About this Gig
Struggling to predict demand accurately across multiple products or locations?
I help businesses forecast sales and demand using Python-based time series models so you can plan inventory and operations with greater confidence instead of guesswork.
What I offer: Data cleaning and exploratory analysis; time series diagnostics and stationarity testing; forecasting with SARIMAX, LightGBM and seasonal models; backtesting and model validation; forecast accuracy metrics such as MAPE; confidence intervals where appropriate; multi-series forecasting across products, stores or regions; clear business-ready reports and dashboards.
Why work with me?
In a recent project, I built a demand forecasting pipeline covering 3,000+ store-department pairs, achieving a 12.6% median MAPE. With a statistics background, I validate assumptions rigorously, compare appropriate approaches, and report honestly including what didn't work.
Tools: Python (statsmodels, LightGBM, pmdarima), SQL and Streamlit.
Let's talk before you order: Send me a brief description of your data, time period, number of series and forecasting goal. I'll review the scope and suggest an appropriate approach.
My Portfolio
FAQ
What data do you need to get started?
Historical sales/demand data (CSV or Excel) with at least 1-2 years of weekly or monthly history works best. More history generally means more accurate forecasts. Not sure if your data is enough? Send a sample and I'll confirm before you order.
How many products or locations can you forecast at once?
Basic and Standard packages cover single or up to 10 series. For larger-scale forecasting (50+ products/stores), check the Premium package, which uses a hybrid multi-model approach to keep both speed and accuracy.
Will I understand the confidence intervals in the forecast?
Yes — every report includes plain-language explanations of forecast accuracy and uncertainty, not just raw numbers, so you can use the results directly for planning decisions.

