How to choose dataset analysis tools
Start with the question you need to answer. Use Dataset Profiler for a file-wide inventory, Correlation Matrix Generator for relationships between numeric columns, Outlier Detector for unusual values, Fuzzy Duplicate Finder for similar records, and Time-Series Anomaly Detector for ordered observations.
How to profile a dataset
Use Dataset Profiler before a narrower check. It accepts CSV, TSV, JSON, JSONL, and spreadsheet files, then reports row and column counts, column types, null and blank rates, unique and duplicate counts, numeric and text statistics, date ranges, top values, and data-quality warnings.
The report can reveal unexpected blanks, repeated values, or a column type that differs from your expectation. Numeric and text statistics show which columns fit a later correlation or outlier check.
How to analyze correlations and find outliers
Choose Correlation Matrix Generator for pairwise relationships between numeric columns. It calculates Pearson, Spearman, or Kendall correlations for CSV, JSON, or spreadsheet input, presents the results in a labelled matrix and heatmap, handles missing values, and exports CSV or JSON.
Choose Outlier Detector for unusually high or low numeric values. Select a method and threshold, optionally group the calculations, then export the flagged observations. A correlation matrix compares columns with each other. An outlier check identifies observations that stand apart within a numeric column or group.
How to find fuzzy duplicates and time-series anomalies
Fuzzy Duplicate Finder finds likely duplicate records in CSV, TSV, JSON, JSONL, or spreadsheet files. Select the columns and configure text similarity, then inspect grouped candidates and choose which record to keep or merge. Nothing is deleted automatically.
Use Time-Series Anomaly Detector for ordered observations in a CSV or JSON file. Choose the timestamp and value columns, handle timezones, duplicates, and gaps, then flag anomalies with a rolling z-score, rolling IQR, or seasonal residual threshold.
Where dataset analysis comes up
- Before a file-wide review, use Dataset Profiler to check column types, blank rates, duplicate counts, statistics, and data-quality warnings.
- When numeric columns need comparison, use Correlation Matrix Generator to calculate a selected correlation method in a labelled matrix and export the result.
- When a numeric column contains unusually high or low values, use Outlier Detector with a chosen method and threshold, then review the flagged observations.
- For likely duplicate records, use Fuzzy Duplicate Finder to group candidates for a keep-or-merge decision. For ordered observations, use Time-Series Anomaly Detector with a timestamp and value column.
When to use Dataset Tools instead
Use Dataset Tools when the job is to clean, reshape, merge, or generate tabular data. Stay in Dataset Analyze Tools when you need to measure an existing file, compare numeric columns, inspect unusual observations, review duplicate candidates, or detect patterns in ordered data.
Frequently asked questions
What is the best dataset analysis tool for a first check?
Start with Dataset Profiler. It reports file dimensions, column types, null and blank rates, duplicate counts, statistics, date ranges, top values, and data-quality warnings.
How do I find duplicate records in a dataset?
Use Fuzzy Duplicate Finder to compare selected columns with configurable text similarity, review grouped candidates, and choose which record to keep or merge.
What file types do Dataset Analyze Tools accept?
Accepted formats vary by tool: Dataset Profiler and Fuzzy Duplicate Finder accept CSV, TSV, JSON, JSONL, and spreadsheet files; Correlation Matrix Generator and Outlier Detector accept CSV, JSON, and spreadsheet files; Time-Series Anomaly Detector accepts CSV and JSON.
How can I check anomalies in time-series data?
Time-Series Anomaly Detector checks ordered CSV or JSON observations after you select timestamp and value columns and choose a rolling z-score, rolling IQR, or seasonal residual threshold.