eTaskHubeTaskHub

Cleaning and Analysis of Retail Sales Dataset (2 Years)

KES 7,500
Data Analysis & Statistics Nairobi Data LabsDeadline: 9/29/2026, 2:00:00 PM

Nairobi Data Labs has been engaged by a mid-sized retail client to clean, structure, and produce an initial exploratory analysis of two years of point-of-sale transaction data covering roughly 40 store locations. The raw dataset, provided as a set of CSV exports, contains known issues including duplicate transaction records, inconsistent date formatting across export batches, missing store identifiers on a subset of rows, and currency values that mix comma and decimal separators depending on the export month. Your task is to first produce a cleaned, standardized version of the dataset with a clear data dictionary explaining every transformation you made and why, so our internal team can audit the process. Once cleaned, we need a descriptive statistical summary covering monthly revenue trends across the full period, identification of the top and bottom performing store locations by both total revenue and revenue per transaction, seasonal patterns (particularly around known regional holiday periods), and any notable anomalies or outliers worth flagging to the client, such as stores with unusual spikes or data gaps. Please use Python (pandas) or R for this work, and deliver both the cleaned dataset and a short written report, no more than four to five pages, summarizing your key findings in plain business language a non-technical retail manager could understand, supported by two to three clear charts. Include your analysis code or notebook so we can verify methodology. We are not looking for predictive modeling on this particular task, just clean data and solid descriptive analysis; depending on quality and turnaround, a follow-up predictive modeling task on the same client relationship may be offered afterward. Strong attention to detail matters more than speed here given the downstream reporting this feeds into, so please flag early if you encounter data issues significant enough to affect the timeline, rather than making silent assumptions. Prior experience cleaning messy real-world retail or transactional data is preferred.

Open for applicationsSign up to apply