LT
LTprophecy
Docs/Data Management/Datasets

Datasets

Datasets are the foundation of your forecasting pipeline. This guide covers everything you need to know about importing, configuring, and managing your data.

Supported Formats

SalesForecaster accepts data in several formats:

  • CSV — Comma-separated values with a header row
  • Excel — .xlsx workbooks (first sheet is used by default)
  • JSON — Array of flat objects
  • API — Push data programmatically via our REST API

Uploading Data

Via the Dashboard

  1. Navigate to Data > Upload
  2. Drag and drop your file or click Browse
  3. Preview the detected schema and adjust column types
  4. Click Save Dataset

Via the API

curl -X POST https://api.ltprophecy.com/v1/datasets \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: multipart/form-data" \
  -F "file=@sales_data.csv" \
  -F "name=Q1 Sales Data"

Data Requirements

Minimum Requirements

  • At least one date column and one numeric target column
  • Minimum 52 data points recommended for weekly data (12 for monthly)
  • Consistent date frequency (no large gaps)

Recommended Structure

For best results, include:

  • Date column — Timestamps for each observation
  • Target variable — The metric to forecast (revenue, units, etc.)
  • Categorical dimensions — Region, product line, customer segment
  • Numeric features — Deal count, marketing spend, headcount

Data Pipeline

The data pipeline automates recurring data imports:

  1. Scheduled pulls — Connect to your data warehouse or CRM
  2. Transformation rules — Apply cleaning and normalization
  3. Validation checks — Automatic data quality monitoring
  4. Incremental updates — Append new data without re-uploading

Configure pipelines under Data > Pipeline.

Feature Engineering

SalesForecaster automatically generates features from your raw data:

  • Temporal features — Day of week, month, quarter, holiday indicators
  • Lag features — Previous period values and rolling averages
  • Growth rates — Period-over-period and year-over-year changes
  • Interaction features — Cross-dimensional aggregations

You can also define custom features under Data > Features.

Data Quality

The platform continuously monitors data quality:

  • Missing value detection — Identifies and flags gaps
  • Outlier detection — Statistical and ML-based anomaly detection
  • Schema drift — Alerts when data structure changes
  • Freshness monitoring — Warns when data hasn't been updated
Getting StartedForecasting