U.S. Granted Patents: Total Patents Originating in California (PATENTUSCATOTAL) | FRED | St. Louis Fed
- Rows
- 29
- Columns
- 2
- Data range
- 1992-01-01 to 2020-01-01
Graph and download economic data for U.S. Granted Patents: Total Patents Originating in California (PATENTUSCATOTAL) from 1992 to 2020 about patent granted, patents, intellectual property, origination, CA, and USA.
AI analysis
Data Analyst Review: U.S. Granted Patents — California Total (PATENTUSCATOTAL)
1. Dataset Overview and Context
This dataset, sourced from the Federal Reserve Bank of St. Louis (FRED) via their public API (fredgraph.csv?id=PATENTUSCATOTAL), tracks the total number of granted U.S. patents originating in California on an annual basis from 1992 to 2020. FRED is a highly reputable, actively maintained data aggregator that pulls this series from USPTO records, making it a trustworthy proxy for innovation activity in the state. Despite being tagged under "Healthcare Prices," the actual content is a general intellectual property/innovation metric — analysts should treat the topic label as a broad categorization rather than a literal description, and validate relevance before pairing it with healthcare-specific studies. Its primary value lies in correlation work examining innovation output against economic, demographic, or industry-specific trends over a nearly three-decade window.
2. Data Quality Observations
The dataset is remarkably clean at face value: zero null cells across both columns, and both observationdate and PATENTUSCATOTAL show 29 distinct values across 29 total rows — indicating no duplicate timestamps and a fully unique, one-record-per-year structure. This is consistent with an annual time series with no missing years in the 1992–2020 span. The duplicate-row count is flagged as pending recomputation ("Phase D — recomputed against Parquet"), so that check should be finalized before final sign-off, but no other quality red flags are apparent. One caveat: the sample data rows are not yet populated, so row-level spot-checking (e.g., verifying date formatting or unexpected value spikes) isn't currently possible — this should be revisited once Parquet wiring is complete.
3. Key Column Distributions
The PATENTUSCATOTAL column ranges from 9,099 to 50,667 patents/year, with a mean of 26,824 and median of 22,200 — the mean exceeding the median suggests a right-skewed distribution, confirmed by the skewness value of 0.479 (moderate positive skew). This aligns with the expected real-world pattern: patent counts were low in the early-to-mid 1990s and grew substantially in later years, pulling the mean upward. The interquartile range (Q1 = 18,855, Q3 = 39,139) spans roughly 20,000 patents, and the standard deviation of 13,222 relative to the mean indicates high variability (~49% coefficient of variation) — typical of a series capturing sustained multi-decade growth rather than a stable equilibrium. This is a classic non-stationary time series, so any correlation analysis should consider differencing, log-transforming, or detrending to avoid spurious correlation driven purely by shared upward trends.
4. Recommended Join Key
The observationdate column (Date type, no nulls, 29 distinct values) is the clear and only sensible join key. Given the annual granularity, cross-dataset joins should aggregate or resample partner datasets to annual frequency (calendar year, matching the YYYY-01-01 format) to align properly. Analysts should confirm whether partner datasets use fiscal years, mid-year snapshots, or different date anchoring before joining.
5. Suggested Pairings for Correlation Analysis
Given the innovation/economic nature of this series, strong correlation candidates include: California-specific GDP or gross state product, R&D spending (federal or private sector), venture capital investment volumes in CA, tech sector employment or wages (BLS/QCEW data), university research funding, and national patent totals (PATENTUSTOTAL, if available) to compute California's share of national innovation output. Since the topic tag references healthcare prices, it may also be worth testing correlations against biotech/pharma patent subcategories or healthcare R&D expenditure series to see if the tagging hints at a deliberate cross-study (e.g., innovation driving healthcare cost/price trends) — though this should be validated rather than assumed given the topic-label/content mismatch noted above.
Columns
- observation_date (date)
- PATENTUSCATOTAL (integer)