What the Houseplant Light Dataset contains
The Houseplant Light Dataset is the machine-readable version of PlantLightIndex's published plant-light records. Each row represents one species, cultivar, hybrid, trade-name entity or horticultural group in the public plant database. The dataset is designed to preserve uncertainty, so a missing practical PPFD or DLI value remains empty instead of being inferred.
The current distribution covers 30 published entities and is generated from the same records that drive the plant profiles, charts, search and calculators.
Measured values, estimates and recommendations are different things
PlantLightIndex treats a measurement as an observation, not automatically as a recommendation. A PPFD reading taken at a leaf tells you how much photosynthetically active photon flux reached that position at that moment. A published greenhouse treatment tells you what researchers supplied under a particular experiment. A photosynthetic light-saturation point describes a physiological response. A practical indoor reference is an editorial recommendation intended to help a houseplant grower make a decision. Those records can inform one another, but they are not interchangeable.
This distinction prevents a common failure in plant-light advice. A plant may survive under a very low PPFD treatment while growing slowly or changing form. Another study may show photosynthesis continuing to increase until a much higher PPFD. Neither number, by itself, proves that the lower value is ideal or that the higher value should be used in a home. PlantLightIndex therefore stores measurement context, evidence scope, source type and recommendation status alongside the number.
Missing evidence stays visible
Some plant profiles contain a practical numerical PPFD range. Others show only a parent-species or genus reference. Some remain qualitative because a defensible species- or cultivar-specific numerical target has not been established in the sources reviewed for the site. That unevenness is intentional. A database becomes less useful when every empty field is filled with a plausible-looking number.
When a value is absent, the site should explain what is known instead. A page may state that a plant is commonly described by an extension source as preferring bright indirect light, or that a cultivar has demonstrated a different physiological response from its green parent, without converting those statements into a precise household target. The calculators then change behavior according to the record: full, contextual or qualitative.
Identity fields
Plant ID and slug provide stable internal and URL identifiers. Display name and primary common name support familiar search language. Accepted scientific name, genus, species and cultivar preserve botanical identity where it is resolved. Entity type distinguishes species from cultivars, hybrids, trade names and horticultural groups.
These identity fields matter because cultivar-level light claims cannot be evaluated correctly if every marketed plant is collapsed into the parent species. The CSV therefore keeps identity separate from recommendation values.
Practical PPFD and contextual PPFD are separate columns
Practical indoor target PPFD is populated only when PlantLightIndex has assigned a numerical practical reference to that record. Contextual PPFD holds a broader reference when it is useful but not exact enough to become the plant's target.
Users analyzing the data should never merge these columns blindly. A contextual range may come from a parent species or genus and is intentionally accompanied by contextual calculator mode. Filtering only for records with a non-null practical range produces the strongest numerical subset.
Physiology fields are not care targets
The dataset can include light-compensation and light-saturation values when those measurements have been captured from research. They describe physiological response under a study context. They must not be substituted for practical indoor targets. A saturation point can sit above, inside or near a practical range without defining the top safe level for a household plant.
This separation allows researchers or advanced users to compare physiology and care interpretation while keeping the public calculator safe.
DLI and photoperiod fields
Target DLI and recommended photoperiod remain null for many records because the evidence needed to prescribe those values is weaker than the evidence for a practical PPFD category. A user can still calculate actual DLI from measured PPFD and hours, but the dataset does not turn that arithmetic into an unsupported plant target.
When future evidence supports a species-specific DLI or photoperiod range, the fields can be populated without changing the data model.
Recommendation mode, basis and status
Calculator mode tells downstream users whether the plant is full, contextual or qualitative. Recommendation basis explains why: direct species evidence, cultivar-specific evidence, extension classification, parent-species reference, genus reference or insufficient quantitative evidence. Recommendation status distinguishes locked, provisional and qualitative-only records.
These columns are essential for any application built on top of the dataset. A numerical range without its basis can look more certain than it is.
Evidence scope and confidence
Evidence scope describes the biological level represented by the evidence: exact cultivar, exact species, species complex, genus, family or general houseplant. Confidence level summarizes the reviewed evidence as High, Medium or Limited, while the internal confidence score remains a decision aid rather than a probability.
For a strict evidence subset, users can filter for full calculator mode plus exact-species/exact-cultivar scope and High confidence. For discovery work, contextual records can be included as long as the broader scope is preserved.
Source identifiers
The CSV lists the source IDs connected to the plant profile. These identifiers correspond to records in the Source Library. They make it possible to inspect provenance without embedding long URLs into every analytical column.
The dataset landing-page schema also lists the underlying source URLs using isBasedOn, following Google's provenance guidance for aggregated datasets. Google's Dataset documentation describes this use of isBasedOn.
Example interpretations
A full record
If a row has a practical PPFD minimum and maximum, full calculator mode and exact-species scope, the range can be used as PlantLightIndex's current practical indoor reference for that species. It still should not be described as a universal optimum or hard damage limit.
A contextual record
If practical PPFD is empty but contextual PPFD is populated, downstream tools should display the range with the evidence scope and basis. It should not be silently copied into the practical target field.
A qualitative record
If both numerical reference fields are empty, the correct interpretation is that no current quantitative target is assigned. The absence itself is information.
How the CSV is generated
The download endpoint reads the version-controlled plant entities and light profiles used by the website, escapes CSV values and streams a current file. This means a plant added to the published dataset can appear in the download as part of the same code release rather than waiting for a separate spreadsheet to be updated.
The build validator checks the relationships before deployment. Full-mode plants must have a practical PPFD range; qualitative plants must not carry a fabricated practical range; source and evidence references must resolve; and range minimums cannot exceed maximums.
Recommended uses
- Compare evidence coverage across plant groups.
- Build private spreadsheets that preserve recommendation mode and confidence.
- Identify cultivars that still lack exact quantitative evidence.
- Join plant records to source IDs for provenance analysis.
- Compare plant practical ranges with measured room PPFD.
- Develop visualizations that clearly separate practical and contextual references.
Do not use the dataset to fabricate missing values or present contextual references as exact care prescriptions.