Forecasting is the primary accountability and has the first call on the Analyst’s time. While the data remediation project runs, transcription and validation will take a large share of the working week, and the forecast must be protected within it. The length of the project depends on how far back the Company chooses to digitise. After handover, the focus shifts to forecasting, keeping the database current, and analysis.
• Produce weekly green leaf forecasts by estate and by field, and rolling monthly forecasts by estate, issued to a fixed weekly deadline and early enough for plucking, transport and factory planning
• Forecast hand-plucked and mechanically harvested leaf separately
• Review and improve the current shoot-count method: sampling design, number and placement of sample plots, and how sample averages convert to kilos per hectare
• Build and maintain forecasting models that combine shoot counts with rainfall, plucking round length, pruning cycle and section age, infill plantings, fertiliser timing, cultivar and seasonal patterns, drawing on the cleaned historical data as it becomes available
• Forecast outgrower leaf volumes with the Outgrower Manager, using historical intake patterns
• Track forecast accuracy weekly and monthly (bias and percentage error against actual leaf received), explain large variances and refine the models, measuring against the Company’s 5% permissible variance standard for crop forecasts
• Prepare crop forecasts for the board and budget cycle when requested
Under the Company’s data project, the Analyst acts as Records and Data Management Officer: he works with the appointed consultant, does the day-to-day data entry and validation, and runs the system once the consultant has left.
• Capture historical records from bound field books, spreadsheets, laboratory reports and soil maps into one central database, using the consultant’s tools and workflow
• Cover the eight data domains of the estate agronomic record: block registry, soil classification maps, soil and leaf analysis, rain gauge logs, fertiliser and other input applications, pruning and weeding, manual harvesting and yield, and mechanical harvester output
• Work in dependency order, from block registry and soil maps, to field condition, to field interventions, to field output, with every record tied to its block ID. No domain is cleaned until the one it depends on has been validated
• Check quality: validate, de-duplicate and consolidate records, verify a 10% stratified sample of every batch against the original source documents, and log and correct errors
• Once the database is live, capture current estate data on a weekly and monthly cycle, set standard capture templates, and train estate clerks and supervisors to use them
• Keep a data dictionary, and maintain the block registry as the single master record of block areas, boundaries, cultivars and section ages, changing it only through a controlled process
• Record infill plantings carried out after each pruning cycle, by block and date, with the number of plants and the cultivar, and tie them to block ID so the block registry stays current
• Take over the database and dashboards at handover using the runbooks supplied, including backups, so the Company has no ongoing dependence on the consultant
• Scope: the nuclear estates’ agronomic data first. Factory processing, auction ledgers, traceability and outgrower records sit outside the current project and may be added later
• Handle estate trial data (for example cultivar, fertiliser, irrigation and harvesting method trials): keep a register of every trial with its objective, layout, treatments, plots and start date, agree standard recording templates with each trial owner, capture trial records on schedule, and tie them to block ID
• Keep trial plots clearly flagged and separate from routine production data, so that trial results are never mixed into, or distort, the forecasts and yield analysis
• Maintain the live dashboard and its views: yield density by block, soil pH alerts, fertiliser and other input cost efficiency, and manual against machine harvesting cost and quality
• Produce standard weekly and monthly reports for management: crop against forecast, budget and prior year, yield per hectare and per field, and plucking productivity
• Flag falls in yield or quality, and soil changes such as acidification, promptly, so corrective action is not delayed by weeks
• Answer ad hoc analysis requests from the General Manager and department managers, for example fertiliser spend against block soil pH, labour productivity against block yield, or the effect of rainfall and pruning on yield
• Present findings clearly to non-technical managers, with plain explanations of the numbers and their reliability
• Analyse trial results with the trial owner: compare treatments using appropriate statistical methods such as analysis of variance, state how confident the result is, and advise whether a trial has enough data to support a decision
• Analyse the effect of infill plantings on stand density, yield and forecast accuracy, and include infill progress in estate performance assessments
Perform any other duties that may be assigned from time to time.
Essential
• Bachelor’s degree in Statistics, Actuarial Science, Data Science, Applied Mathematics, Economics, Computer Science or a closely related quantitative field from a recognised university
• At least 3 years’ experience in forecasting, data analysis or data management, in any sector
• Demonstrated experience building forecasts or predictive models from real operational data, and measuring their accuracy
• Working knowledge of SQL and relational databases, and advanced Excel
• Experience building dashboards or reports; experience in tools such as Power BI, Tableau, Superset or Metabase is helpful
• Proven accuracy and patience in detailed data entry, data cleaning and checking, including from handwritten or scanned records
• Willingness to be based on an estate and to spend time in the field and in estate record rooms
Desirable
• Master’s degree in a quantitative field
• Experience on a records digitisation or data migration project
• Experience with PostgreSQL, and with GIS or mapping tools such as PostGIS
• Experience with time series methods and with sampling design
• Experience in agriculture, agribusiness, commodities, supply chain or manufacturing planning
• Knowledge of tea or other perennial crops such as coffee, sugar or avocado
• Experience in experimental design and the statistical analysis of field trials
• Statistical and forecasting ability: understands sampling, variance, regression and time series, and can say how reliable a forecast is
• Data discipline: careful and methodical, checks every number, and documents how each figure was produced
• Accuracy under repetition: keeps quality high through long stretches of detailed transcription and verification
• Systems thinking: understands how block, soil, rainfall, input, pruning and yield data connect, and keeps those links intact
• Communication: explains technical results simply to managers and supervisors, in writing and in person
• Training and influence: respectful and practical when asking estate staff to change how they record data, and able to train them and earn their cooperation
• Reliability: delivers the weekly forecast on time, every week, including under pressure
• Curiosity: willing to learn how tea is grown and plucked in order to make the models sensible
• Confidentiality: handles commercial and production data with discretion