Source: /home/coolhand/html/datavis/data_trove/entertainment/gaming/recommendations_metadata.json
Saturn profiled 1 rows across 6 columns. The stats below are deterministic and machine-readable; the prose is a language-model interpretation of those stats (opt-in, added after the fact, never sees raw rows).
!pip install saturn-dissect
import subprocess
subprocess.run([
"saturn", "analyze", "/home/coolhand/html/datavis/data_trove/entertainment/gaming/recommendations_metadata.json",
"--findings", "data-trove-steam-user-reviews.json",
"--llm", "anthropic:default",
])This is a single-row metadata descriptor for the 'Steam Game Recommendations' dataset, last updated 2025-01-20 — it is a catalog entry rather than the underlying data itself. The key takeaway is the scale of what it describes: 41.1 million Steam user reviews stored in a 1.9 GB file, sourced likely via Kaggle or SteamSpy. The metadata notes that the full dataset links to companion files (games.csv and users.csv) via app_id and user_id, and includes playtime and helpfulness metrics — making those join keys the most important fields to validate before any analysis. Analysts should treat this file as a data dictionary and move quickly to the referenced source files for substantive exploration.
citing: record_count.max · dataset_name.top_value · last_updated.top_value · source.top_value · notes.top_value · row_count
saturn.schema() · 6 columns
| column | kind | n | null% | unique | alerts |
|---|---|---|---|---|---|
| dataset_name | categorical | 1 | 0.0% | 1 | long_tail imbalance |
| last_updated | categorical | 1 | 0.0% | 1 | long_tail imbalance |
| source | categorical | 1 | 0.0% | 1 | long_tail imbalance |
| record_count | numeric | 1 | 0.0% | 1 | constant |
| fields | unknown | 1 | 0.0% | — | skipped |
| notes | categorical | 1 | 0.0% | 1 | long_tail imbalance |
| value | count | share |
|---|---|---|
| Steam Game Recommendations | 1 | 100.0% |
| bin | count |
|---|---|
| 4.115e+07 – 4.115e+07 | 0 |
| 4.115e+07 – 4.115e+07 | 0 |
| 4.115e+07 – 4.115e+07 | 1 |
| 4.115e+07 – 4.115e+07 | 0 |
| 4.115e+07 – 4.115e+07 | 0 |
| value | count | share |
|---|---|---|
| Steam Store user reviews (likely via Kaggle or SteamSpy) | 1 | 100.0% |
| value | count | share |
|---|---|---|
| 2025-01-20 | 1 | 100.0% |
| column | kind | null % |
|---|---|---|
| dataset_name | categorical | 0.0% |
| last_updated | categorical | 0.0% |
| source | categorical | 0.0% |
| record_count | numeric | 0.0% |
| fields | unknown | 0.0% |
| notes | categorical | 0.0% |
This column is a dataset-level metadata tag identifying the source dataset, with every single row (n=1) carrying the value 'Steam Game Recommendations'. Cardinality is 1 and entropy is 0.0, meaning the column is entirely constant and carries zero information. The 'long_tail' and 'imbalance' alerts are triggered mechanically by the extreme top_rate of 1.0, but are not meaningful here — this is simply a constant label.
Treatment: Drop before modelling; zero-variance constant column adds no predictive signal.
saturn.columns["dataset_name"].stats
| stat | value |
|---|---|
| n | 1 |
| nulls | 0 (0.0%) |
| unique | 1 |
| top_value | Steam Game Recommendations |
| top_rate | 1 |
| cardinality | 1 |
| entropy | 0 |
| entropy_ratio | 0 |
| alert: long_tail | 1 singleton categories |
| alert: imbalance | top value is 100.0% of rows |
dataset_name.
| value | count | share |
|---|---|---|
| Steam Game Recommendations | 1 | 100.0% |
This column is a metadata timestamp recording when each record was last updated. With only 1 row in the dataset and a single value of '2025-01-20' holding a top_rate of 1.0, the column is entirely constant — there is zero variance. The alerts for long_tail and imbalance are technically correct but vacuous given the single-row dataset; no meaningful distribution analysis is possible.
Treatment: Exclude from modelling features; retain as audit metadata, but re-evaluate once the full dataset is loaded.
saturn.columns["last_updated"].stats
| stat | value |
|---|---|
| n | 1 |
| nulls | 0 (0.0%) |
| unique | 1 |
| top_value | 2025-01-20 |
| top_rate | 1 |
| cardinality | 1 |
| entropy | 0 |
| entropy_ratio | 0 |
| alert: long_tail | 1 singleton categories |
| alert: imbalance | top value is 100.0% of rows |
last_updated.
| value | count | share |
|---|---|---|
| 2025-01-20 | 1 | 100.0% |
This column records the data provenance/source attribution for the dataset, and contains exactly one unique value across all rows: 'Steam Store user reviews (likely via Kaggle or SteamSpy)'. With cardinality of 1, entropy of 0.0, and a top_rate of 1.0, it is a constant column carrying zero discriminative information. The alerts for long_tail and imbalance are technically triggered but are trivially explained by the single-value nature of the column.
Treatment: Drop before modelling; constant column adds no signal and wastes memory.
saturn.columns["source"].stats
| stat | value |
|---|---|
| n | 1 |
| nulls | 0 (0.0%) |
| unique | 1 |
| top_value | Steam Store user reviews (likely via Kaggle or SteamSpy) |
| top_rate | 1 |
| cardinality | 1 |
| entropy | 0 |
| entropy_ratio | 0 |
| alert: long_tail | 1 singleton categories |
| alert: imbalance | top value is 100.0% of rows |
source.
| value | count | share |
|---|---|---|
| Steam Store user reviews (likely via Kaggle or SteamSpy) | 1 | 100.0% |
This column appears to be a summary or metadata field recording total row count for a dataset or batch, with a single observed value of 41,154,794. The dataset profile contains only 1 row (n=1), meaning this column is a scalar summary rather than a per-record attribute. It is flagged as 'constant' with zero variance, zero IQR, and min/max/mean/median all equal to 41,154,794.0. There is no analytical signal here — it carries no discriminative power and exists purely as a metadata annotation.
Treatment: Drop before modelling; retain only as a data-quality audit reference if needed.
saturn.columns["record_count"].stats
| stat | value |
|---|---|
| n | 1 |
| nulls | 0 (0.0%) |
| unique | 1 |
| min | 4.115e+07 |
| max | 4.115e+07 |
| mean | 4.115e+07 |
| median | 4.115e+07 |
| std | 0 |
| q1 | 4.115e+07 |
| q3 | 4.115e+07 |
| iqr | 0 |
| skew | 0 |
| kurtosis | 0 |
| n_outliers | 0 |
| outlier_rate | 0 |
| zero_rate | 0 |
| alert: constant | only one distinct value |
record_count. Vertical dash marks the median.
| bin | count |
|---|---|
| 4.115e+07 – 4.115e+07 | 0 |
| 4.115e+07 – 4.115e+07 | 0 |
| 4.115e+07 – 4.115e+07 | 1 |
| 4.115e+07 – 4.115e+07 | 0 |
| 4.115e+07 – 4.115e+07 | 0 |
This column ('fields') contains only a single row and was skipped by the profiler, yielding no distributional statistics. With n=1 and no type inference completed, essentially nothing can be determined about its content or role. The absence of nulls is the only positive signal available.
Treatment: Inspect raw value manually before deciding on handling; re-profile with a larger sample if this column appears in a fuller dataset.
saturn.columns["fields"].stats
| stat | value |
|---|---|
| n | 1 |
| nulls | 0 (0.0%) |
| unique | — |
| alert: skipped | no profiler for kind=unknown |
This column is a dataset-level metadata note, not a real data column — its single value is a documentation string describing the broader dataset (41.1 million Steam reviews, file size, join keys, and available metrics). With n=1, cardinality=1, and top_rate=1.0, it carries zero analytical signal and is purely an artifact of how the dataset profile was constructed. The entropy of 0.0 confirms there is no variation whatsoever.
Treatment: Drop before modelling; this is a documentation artifact with no predictive or analytical value.
saturn.columns["notes"].stats
| stat | value |
|---|---|
| n | 1 |
| nulls | 0 (0.0%) |
| unique | 1 |
| top_value | 41.1 million Steam user reviews/recommendations. File is 1.9 GB. Links to games.csv via app_id and to users.csv via user_id. Includes playtime and helpfulness metrics. |
| top_rate | 1 |
| cardinality | 1 |
| entropy | 0 |
| entropy_ratio | 0 |
| alert: long_tail | 1 singleton categories |
| alert: imbalance | top value is 100.0% of rows |
notes.
| value | count | share |
|---|---|---|
| 41.1 million Steam user reviews/recommendations. File is 1.9 GB. Links to games.csv via app_id and to users.csv via user_id. Includes playtime and helpfulness metrics. | 1 | 100.0% |
click to copy
@misc{saturn-data-trove-steam-user-reviews-2026,
author = {Steuber, Luke},
title = {Saturn reading: data trove steam user reviews},
year ={2026},
howpublished = {\url{https://dr.eamer.dev/saturn/view/data-trove-steam-user-reviews}},
note = {Profiled with saturn-dissect v0.2.0, prompt saturn-insight-v2, model anthropic:default},
}
Steuber, L. (2026). Saturn reading: data trove steam user reviews. Source: /home/coolhand/html/datavis/data_trove/entertainment/gaming/recommendations_metadata.json. Profiled with saturn-dissect v0.2.0 (saturn-insight-v2, anthropic:default). Retrieved from https://dr.eamer.dev/saturn/view/data-trove-steam-user-reviews