[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M79VMXXN4T5XDT99A8YA4MBR":3,"skill-same:01M79VMXXN4T5XDT99A8YA4MBR":48},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":29,"ownerHtmlUrl":30,"ownerBlogUrl":31,"ownerDescription":32,"updatedAt":33,"install":34,"resources":47},"01M79VMXXN4T5XDT99A8YA4MBR","python-data",null,"Performs heavy numerical compute, statistical analysis, and machine learning inside the sandbox. Joins\u002Faggregates tables to answer the question and (when warranted) builds models — saving every result table downstream skills chart and report.","Vishwajeetsrk\u002FJARVIS-AI-OS","Vishwajeetsrk","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F83073825?v=4","01M79V4YAYQ35CPZPBS1RQ6VEX","VishwaJeet","01M79V4YBP8A6PPPZ91229KVHN",11,"MIT","https:\u002F\u002Fgithub.com\u002FVishwajeetsrk\u002FJARVIS-AI-OS\u002Fblob\u002Fmain\u002FProjects\u002Fai-data-analyst\u002Fagent\u002Fskills\u002Fpython-data\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: python-data\ndescription: >-\n  Performs heavy numerical compute, statistical analysis, and machine learning\n  inside the sandbox. Joins\u002Faggregates tables to answer the question and (when\n  warranted) builds models — saving every result table downstream skills chart\n  and report.\n---\n\n# Python data skill\n\nRun complex calculations, aggregations, statistical tests, regressions, and ML\nusing `pandas`, `numpy`, and `scikit-learn` in the sandbox. This is\n**agent-driven**: you write Python and run it with the `code_execution` tool —\nthere is no fixed script. Ground every computation in\n`.\u002Fworkspace\u002Fdata\u002Fprofile.json` (written by the `data-explorer` skill).\n\n## Workflow\n\n1. **Read the profile.** Load `.\u002Fworkspace\u002Fdata\u002Fprofile.json` to know each\n   column's `inferred_type`, the measures vs dimensions, and any `join_keys`.\n2. **Write a Python script** using `pandas`, `numpy`, or `sklearn` to:\n   - Join, filter, and aggregate tables (use the detected join keys).\n   - Compute statistics (mean, median, stddev, correlations).\n   - Run statistical tests (chi-square, t-test) when appropriate.\n   - Build or evaluate ML models when the question warrants it (see Modeling).\n3. **Execute in the sandbox.** Run via `python3 script.py`.\n4. **Handle errors.** If the code fails, read the traceback and fix it.\n5. **Save result tables** to `.\u002Fworkspace\u002Fdata\u002Fanalysis\u002F*.csv` (see Output\n   contract) and print key values to the run log.\n\n## Loading and coercing data\n\n```python\nimport pandas as pd, numpy as np, os, json\n\nWS = \".\u002Fworkspace\"\nwith open(os.path.join(WS, \"data\u002Fprofile.json\")) as f:\n    profile = json.load(f)\n\ndef coerce(df, cols):\n    for col in cols:\n        name, t = col[\"name\"], col[\"inferred_type\"]\n        if name not in df.columns:\n            continue\n        if t == \"numeric\" and df[name].dtype == object:\n            df[name] = pd.to_numeric(\n                df[name].astype(str).str.replace(r\"[,$£€%\\s]\", \"\", regex=True),\n                errors=\"coerce\",\n            )\n        elif t == \"datetime\":\n            try:\n                # Cast to string first to prevent list\u002Fobject errors from crashing to_datetime\n                df[name] = pd.to_datetime(df[name].astype(str), errors=\"coerce\")\n            except Exception:\n                pass\n    return df\n\ntables = {}\nfor t in profile.get(\"tables\", []):\n    path = os.path.join(WS, \"data\", t[\"file\"])\n    tables[t[\"name\"]] = coerce(pd.read_csv(path, sep=None, engine=\"python\"), t[\"columns\"])\n```\n\n## Common patterns\n\n- **Revenue (line-item)**: Northwind has no revenue column — derive it from order\n  line items before aggregating:\n  `od[\"revenue\"] = od[\"unitPrice\"] * od[\"quantity\"] * (1 - od[\"discount\"])`.\n  This excludes freight; add `orders[\"freight\"]` separately only if the question\n  asks for total spend including shipping.\n- **Aggregation**: `df.groupby(\"category\")[\"revenue\"].sum().sort_values(ascending=False)`\n- **Top-N**: `.head(10)` after sorting.\n- **Time series**: set a datetime index and `resample(\"ME\")` \u002F `\"W\"` \u002F `\"QE\"`, then `.sum()` \u002F `.mean()`.\n- **Growth**: `series.pct_change()`; cumulative with `.cumsum()`.\n- **Pivot**: `df.pivot_table(index=\"region\", columns=\"category\", values=\"revenue\", aggfunc=\"sum\")`.\n- **Cohort \u002F RFM**: bucket customers by Recency, Frequency, Monetary; score with `pd.qcut`.\n- **Correlation**: `df[numeric_cols].corr()`.\n- **Joining relational tables**: merge on `join_keys` before aggregating, e.g.\n  `order_details.merge(products[[\"productID\",\"categoryID\"]], on=\"productID\")`.\n  Drop junk\u002FBLOB columns (e.g. `categories.picture`) before joining.\n\n## Modeling (only when warranted)\n\nUse ML **only when the question calls for it** — forecasting, prediction,\nsegmentation, or driver analysis. For purely descriptive questions, skip it.\n\n| Question shape | Approach | scikit-learn |\n|----------------|----------|--------------|\n| \"Forecast next period\" | Time-series regression on time features | `LinearRegression`, `RandomForestRegressor` |\n| \"Predict a numeric outcome\" | Regression | `Ridge`, `RandomForestRegressor` |\n| \"Predict a category \u002F churn\" | Classification | `LogisticRegression`, `RandomForestClassifier` |\n| \"Group similar customers\" | Clustering | `KMeans` + `StandardScaler` |\n| \"What drives this metric?\" | Feature importance \u002F coefficients | tree `feature_importances_`, linear `coef_` |\n\n- Split with `train_test_split(random_state=42)`; for time series split\n  chronologically (do NOT shuffle).\n- Always **evaluate and report** the metric: regression → R²\u002FMAE\u002FRMSE;\n  classification → accuracy\u002Fprecision\u002Frecall\u002FROC-AUC; clustering → silhouette.\n- Be honest about uncertainty, small samples, and leakage. Never present a\n  forecast as certainty.\n\n## Output contract — `.\u002Fworkspace\u002Fdata\u002Fanalysis\u002F*.csv`\n\n> [!IMPORTANT]\n> The `visualization` skill (`make_chart.py --data ...`) and the `reporting`\n> skill (`build_report.py`) read every CSV in `.\u002Fworkspace\u002Fdata\u002Fanalysis\u002F`.\n> Save each table you intend to chart or report there. Keep them small and tidy\n> (already aggregated, typically ≤ 20 rows for display tables).\n\n```python\nout = os.path.join(WS, \"data\u002Fanalysis\")\nos.makedirs(out, exist_ok=True)\n\nrev_by_cat = df.groupby(\"category\")[\"revenue\"].sum().sort_values(ascending=False).reset_index()\nrev_by_cat.to_csv(os.path.join(out, \"revenue_by_category.csv\"), index=False)\n\nmonthly = df.set_index(\"order_date\")[\"revenue\"].resample(\"ME\").sum().reset_index()\nmonthly.to_csv(os.path.join(out, \"revenue_monthly.csv\"), index=False)\n```\n\nSuggested names for model outputs (so they chart cleanly downstream):\n\n- Forecasts → `data\u002Fanalysis\u002Fforecast.csv` (date, actual, predicted)\n- Segments → `data\u002Fanalysis\u002Fsegments.csv` (segment id + profile metrics)\n- Drivers → `data\u002Fanalysis\u002Ffeature_importance.csv` (feature, importance)\n\n## Notes\n\n- Print the result of every computation so values are captured in the run log.\n- Handle NaNs explicitly (`dropna`, `fillna`) and state what you did.\n- Note the date range and row counts behind each aggregate.\n- Use `select_dtypes(include=[\"object\", \"str\"])` for categorical columns\n  (avoids the pandas 4 deprecation warning).\n- Install missing packages with `pip install -q \u003Cpackage>` before importing.\n- Never invent numbers — if a needed column does not exist, say so and adjust.\n","# Python data skill\n\nRun complex calculations, aggregations, statistical tests, regressions, and ML\nusing `pandas`, `numpy`, and `scikit-learn` in the sandbox. This is\n**agent-driven**: you write Python and run it with the `code_execution` tool —\nthere is no fixed script. Ground every computation in\n`.\u002Fworkspace\u002Fdata\u002Fprofile.json` (written by the `data-explorer` skill).\n\n## Workflow\n\n1. **Read the profile.** Load `.\u002Fworkspace\u002Fdata\u002Fprofile.json` to know each\n   column's `inferred_type`, the measures vs dimensions, and any `join_keys`.\n2. **Write a Python script** using `pandas`, `numpy`, or `sklearn` to:\n   - Join, filter, and aggregate tables (use the detected join keys).\n   - Compute statistics (mean, median, stddev, correlations).\n   - Run statistical tests (chi-square, t-test) when appropriate.\n   - Build or evaluate ML models when the question warrants it (see Modeling).\n3. **Execute in the sandbox.** Run via `python3 script.py`.\n4. **Handle errors.** If the code fails, read the traceback and fix it.\n5. **Save result tables** to `.\u002Fworkspace\u002Fdata\u002Fanalysis\u002F*.csv` (see Output\n   contract) and print key values to the run log.\n\n## Loading and coercing data\n\n```python\nimport pandas as pd, numpy as np, os, json\n\nWS = \".\u002Fworkspace\"\nwith open(os.path.join(WS, \"data\u002Fprofile.json\")) as f:\n    profile = json.load(f)\n\ndef coerce(df, cols):\n    for col in cols:\n        name, t = col[\"name\"], col[\"inferred_type\"]\n        if name not in df.columns:\n            continue\n        if t == \"numeric\" and df[name].dtype == object:\n            df[name] = pd.to_numeric(\n                df[name].astype(str).str.replace(r\"[,$£€%\\s]\", \"\", regex=True),\n                errors=\"coerce\",\n            )\n        elif t == \"datetime\":\n            try:\n                # Cast to string first to prevent list\u002Fobject errors from crashing to_datetime\n                df[name] = pd.to_datetime(df[name].astype(str), errors=\"coerce\")\n            except Exception:\n                pass\n    return df\n\ntables = {}\nfor t in profile.get(\"tables\", []):\n    path = os.path.join(WS, \"data\", t[\"file\"])\n    tables[t[\"name\"]] = coerce(pd.read_csv(path, sep=None, engine=\"python\"), t[\"columns\"])\n```\n\n## Common patterns\n\n- **Revenue (line-item)**: Northwind has no revenue column — derive it from order\n  line items before aggregating:\n  `od[\"revenue\"] = od[\"unitPrice\"] * od[\"quantity\"] * (1 - od[\"discount\"])`.\n  This excludes freight; add `orders[\"freight\"]` separately only if the question\n  asks for total spend including shipping.\n- **Aggregation**: `df.groupby(\"category\")[\"revenue\"].sum().sort_values(ascending=False)`\n- **Top-N**: `.head(10)` after sorting.\n- **Time series**: set a datetime index and `resample(\"ME\")` \u002F `\"W\"` \u002F `\"QE\"`, then `.sum()` \u002F `.mean()`.\n- **Growth**: `series.pct_change()`; cumulative with `.cumsum()`.\n- **Pivot**: `df.pivot_table(index=\"region\", columns=\"category\", values=\"revenue\", aggfunc=\"sum\")`.\n- **Cohort \u002F RFM**: bucket customers by Recency, Frequency, Monetary; score with `pd.qcut`.\n- **Correlation**: `df[numeric_cols].corr()`.\n- **Joining relational tables**: merge on `join_keys` before aggregating, e.g.\n  `order_details.merge(products[[\"productID\",\"categoryID\"]], on=\"productID\")`.\n  Drop junk\u002FBLOB columns (e.g. `categories.picture`) before joining.\n\n## Modeling (only when warranted)\n\nUse ML **only when the question calls for it** — forecasting, prediction,\nsegmentation, or driver analysis. For purely descriptive questions, skip it.\n\n| Question shape | Approach | scikit-learn |\n|----------------|----------|--------------|\n| \"Forecast next period\" | Time-series regression on time features | `LinearRegression`, `RandomForestRegressor` |\n| \"Predict a numeric outcome\" | Regression | `Ridge`, `RandomForestRegressor` |\n| \"Predict a category \u002F churn\" | Classification | `LogisticRegression`, `RandomForestClassifier` |\n| \"Group similar customers\" | Clustering | `KMeans` + `StandardScaler` |\n| \"What drives this metric?\" | Feature importance \u002F coefficients | tree `feature_importances_`, linear `coef_` |\n\n- Split with `train_test_split(random_state=42)`; for time series split\n  chronologically (do NOT shuffle).\n- Always **evaluate and report** the metric: regression → R²\u002FMAE\u002FRMSE;\n  classification → accuracy\u002Fprecision\u002Frecall\u002FROC-AUC; clustering → silhouette.\n- Be honest about uncertainty, small samples, and leakage. Never present a\n  forecast as certainty.\n\n## Output contract — `.\u002Fworkspace\u002Fdata\u002Fanalysis\u002F*.csv`\n\n> [!IMPORTANT]\n> The `visualization` skill (`make_chart.py --data ...`) and the `reporting`\n> skill (`build_report.py`) read every CSV in `.\u002Fworkspace\u002Fdata\u002Fanalysis\u002F`.\n> Save each table you intend to chart or report there. Keep them small and tidy\n> (already aggregated, typically ≤ 20 rows for display tables).\n\n```python\nout = os.path.join(WS, \"data\u002Fanalysis\")\nos.makedirs(out, exist_ok=True)\n\nrev_by_cat = df.groupby(\"category\")[\"revenue\"].sum().sort_values(ascending=False).reset_index()\nrev_by_cat.to_csv(os.path.join(out, \"revenue_by_category.csv\"), index=False)\n\nmonthly = df.set_index(\"order_date\")[\"revenue\"].resample(\"ME\").sum().reset_index()\nmonthly.to_csv(os.path.join(out, \"revenue_monthly.csv\"), index=False)\n```\n\nSuggested names for model outputs (so they chart cleanly downstream):\n\n- Forecasts → `data\u002Fanalysis\u002Fforecast.csv` (date, actual, predicted)\n- Segments → `data\u002Fanalysis\u002Fsegments.csv` (segment id + profile metrics)\n- Drivers → `data\u002Fanalysis\u002Ffeature_importance.csv` (feature, importance)\n\n## Notes\n\n- Print the result of every computation so values are captured in the run log.\n- Handle NaNs explicitly (`dropna`, `fillna`) and state what you did.\n- Note the date range and row counts behind each aggregate.\n- Use `select_dtypes(include=[\"object\", \"str\"])` for categorical columns\n  (avoids the pandas 4 deprecation warning).\n- Install missing packages with `pip install -q \u003Cpackage>` before importing.\n- Never invent numbers — if a needed column does not exist, say so and adjust.\n",{"name":5,"description":7},"main","87476a719bbe41680362bb9bfacf884c53784783","Projects\u002Fai-data-analyst\u002Fagent\u002Fskills\u002Fpython-data",3,"https:\u002F\u002Fgithub.com\u002FVishwajeetsrk\u002FJARVIS-AI-OS","https:\u002F\u002Fjarvisaios.vercel.app","https:\u002F\u002Fgithub.com\u002FVishwajeetsrk","https:\u002F\u002Flearnifyaitool.vercel.app\u002F","🎉 Welcome to my profile! I'm a BCA student with a passion for technology and innovation. As a frontend developer.","2026-09-01 14:20:58",{"source":8,"skillName":5,"commands":35,"prompt":45,"zipAvailable":46},[36,39,42],{"tool":37,"command":38},"npx","npx skills add Vishwajeetsrk\u002FJARVIS-AI-OS --skill python-data",{"tool":40,"command":41},"bunx","bunx skills add Vishwajeetsrk\u002FJARVIS-AI-OS --skill python-data",{"tool":43,"command":44},"pnpm","pnpm dlx skills add Vishwajeetsrk\u002FJARVIS-AI-OS --skill python-data","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FVishwajeetsrk\u002FJARVIS-AI-OS\u002Fblob\u002Fmain\u002FProjects\u002Fai-data-analyst\u002Fagent\u002Fskills\u002Fpython-data\u002FSKILL.md",true,[],[49,55,64,72,78,86],{"skillNo":50,"name":51,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":52,"categories":53,"keywords":54,"publishedAt":19},"01M79VY332PY7B4KCWMN3FM430","workspace-agent","https:\u002F\u002Fgithub.com\u002FVishwajeetsrk\u002FJARVIS-AI-OS\u002Fblob\u002Fmain\u002Fskills\u002Fworkspace-agent\u002FSKILL.md",[],[],{"skillNo":56,"name":57,"nameZh":6,"description":58,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":59,"licenseFileUrl":60,"sourceHtmlUrl":61,"categories":62,"keywords":63,"publishedAt":19},"01M79VXZK9MKMYDV57N8VJDPYW","webapp-testing","Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs.","Apache-2.0","https:\u002F\u002Fgithub.com\u002FVishwajeetsrk\u002FJARVIS-AI-OS\u002Fblob\u002Fmain\u002Fskills\u002Fwebapp-testing\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002FVishwajeetsrk\u002FJARVIS-AI-OS\u002Fblob\u002Fmain\u002Fskills\u002Fwebapp-testing\u002FSKILL.md",[],[],{"skillNo":65,"name":66,"nameZh":6,"description":67,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":59,"licenseFileUrl":68,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":19},"01M79VXR975D5MQTW5VMFTKWGG","web-artifacts-builder","Suite of tools for creating elaborate, multi-component claude.ai HTML artifacts using modern frontend web technologies (React, Tailwind CSS, shadcn\u002Fui). Use for complex artifacts requiring state management, routing, or shadcn\u002Fui components - not for simple single-file HTML\u002FJSX artifacts.","https:\u002F\u002Fgithub.com\u002FVishwajeetsrk\u002FJARVIS-AI-OS\u002Fblob\u002Fmain\u002Fskills\u002Fweb-artifacts-builder\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002FVishwajeetsrk\u002FJARVIS-AI-OS\u002Fblob\u002Fmain\u002Fskills\u002Fweb-artifacts-builder\u002FSKILL.md",[],[],{"skillNo":73,"name":74,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":75,"categories":76,"keywords":77,"publishedAt":19},"01M79VXMF2JMWE9GETMJFRM4B9","ui-ux-expert","https:\u002F\u002Fgithub.com\u002FVishwajeetsrk\u002FJARVIS-AI-OS\u002Fblob\u002Fmain\u002Fskills\u002Fui-ux-expert\u002FSKILL.md",[],[],{"skillNo":79,"name":80,"nameZh":6,"description":81,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":59,"licenseFileUrl":82,"sourceHtmlUrl":83,"categories":84,"keywords":85,"publishedAt":19},"01M79VXGG2RRFF3EH79CDYVE0K","theme-factory","Toolkit for styling artifacts with a theme. These artifacts can be slides, docs, reportings, HTML landing pages, etc. There are 10 pre-set themes with colors\u002Ffonts that you can apply to any artifact that has been creating, or can generate a new theme on-the-fly.","https:\u002F\u002Fgithub.com\u002FVishwajeetsrk\u002FJARVIS-AI-OS\u002Fblob\u002Fmain\u002Fskills\u002Ftheme-factory\u002FLICENSE.txt","https:\u002F\u002Fgithub.com\u002FVishwajeetsrk\u002FJARVIS-AI-OS\u002Fblob\u002Fmain\u002Fskills\u002Ftheme-factory\u002FSKILL.md",[],[],{"skillNo":87,"name":88,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":89,"categories":90,"keywords":91,"publishedAt":19},"01M79VXCR3A6QMYT131RFKNTWT","test-agent","https:\u002F\u002Fgithub.com\u002FVishwajeetsrk\u002FJARVIS-AI-OS\u002Fblob\u002Fmain\u002Fskills\u002Ftest-agent\u002FSKILL.md",[],[]]