{ "slug": "one-hot-encoder", "name": "gizza-ai/one-hot-encoder", "version": "0.1.0", "title": "One-Hot Encoder — Expand a CSV Column Into Dummy Variables — gizza.ai", "description": "One-hot encode a CSV categorical column into binary 0/1 indicator columns. Drop a reference level, cap top categories, and keep or remove the original.", "tags": [ "one-hot encoding", "dummy variables", "categorical encoding", "get_dummies", "feature engineering", "csv encoder", "machine learning preprocessing", "indicator variables" ], "category": "encoding", "urls": { "page": "https://gizza.ai/tools/one-hot-encoder/", "markdown": "https://gizza.ai/tools/one-hot-encoder/index.md", "descriptor": "https://gizza.ai/tools/one-hot-encoder/tool.json", "deep_link_example": "https://gizza.ai/tools/one-hot-encoder/?data=city%2Cn%0AParis%2C1%0ARome%2C2%0AParis%2C3&column=city&prefix=city&separator=_&drop=none&drop_original=true&missing=zeros&max_categories=0&min_count=0&other_column=true&positive=1&negative=0&case_sensitive=true&sort=alphabetical&has_header=true&delimiter=comma" }, "cli": "gizza tool one-hot-encoder \"city,n\nParis,1\nRome,2\nParis,3\" 'column=city'", "tool": { "description": "Expand a categorical CSV column into one binary indicator column per distinct category (one-hot / dummy-variable encoding, as produced by pandas get_dummies or scikit-learn OneHotEncoder). Each row gets a 1 in the column matching its value and 0 elsewhere. Supports a custom column prefix and separator, dropping a reference level (first / last / only-if-binary) to avoid the dummy-variable trap, keeping or removing the source column, capping the expansion to the top-N most frequent categories or to values seen at least N times with an optional combined 'other' column, custom positive/negative values (1/0, true/false), blank-cell handling (zeros / separate NaN column / blank / error), case-insensitive grouping, alphabetical / frequency / first-seen column ordering, header or 1-based index column selection, and comma/tab/semicolon/pipe delimiters.", "parameters": { "additionalProperties": false, "properties": { "case_sensitive": { "default": true, "description": "Give values differing only in case their own columns. Turn off to fold 'Paris', 'PARIS', and 'paris' into one column named after the first spelling seen.", "type": "boolean" }, "column": { "description": "Categorical column to expand: a header name, or a 1-based column number when there is no header. Example: city.", "type": "string" }, "data": { "description": "CSV text to encode. The chosen column is expanded into one binary indicator (dummy) column per distinct value.", "type": "string" }, "delimiter": { "default": "comma", "description": "CSV delimiter used to read and write the data.", "enum": [ "comma", "tab", "semicolon", "pipe" ], "type": "string" }, "drop": { "default": "none", "description": "Drop one category as the reference level to avoid the dummy-variable trap (perfect collinearity in linear models): 'none' keeps all k columns, 'first'/'last' keep k-1, 'if-binary' drops the first level only when the column has exactly two categories.", "enum": [ "none", "first", "last", "if-binary" ], "type": "string" }, "drop_original": { "default": true, "description": "Remove the original categorical column from the output, keeping only the indicator columns. Turn off to keep it alongside them.", "type": "boolean" }, "has_header": { "default": true, "description": "Treat the first CSV row as headers. Turn off to select the column by 1-based number; no header row is then written.", "type": "boolean" }, "max_categories": { "default": 0, "description": "Keep only the N most frequent categories and give the rest no column of their own. 0 keeps every category. Use this on high-cardinality columns to stop the output exploding.", "maximum": 512, "minimum": 0, "type": "integer" }, "min_count": { "default": 0, "description": "Keep only categories occurring at least this many times, so one-off values do not each get a column. 0 or 1 keeps every category.", "maximum": 100000, "minimum": 0, "type": "integer" }, "missing": { "default": "zeros", "description": "What a blank cell means: 'zeros' writes 0 in every indicator, 'separate' adds its own _NaN indicator column, 'blank' leaves the indicator cells empty, 'error' rejects the input instead of guessing.", "enum": [ "zeros", "separate", "blank", "error" ], "type": "string" }, "negative": { "default": "0", "description": "Text written when a row does not belong to that category. Default '0'; use 'false' or 'N' to match the positive value's style.", "type": "string" }, "other_column": { "default": false, "description": "Add one combined _other indicator for the categories excluded by max_categories or min_count. Off means those rows are 0 in every column.", "type": "boolean" }, "positive": { "default": "1", "description": "Text written when a row belongs to that category. Default '1'; use 'true' or 'Y' for a boolean-style output.", "type": "string" }, "prefix": { "description": "Prefix for the generated column names. Leave blank to use the source column's own name, so 'city' produces city_Paris, city_Rome.", "type": "string" }, "separator": { "default": "_", "description": "Text placed between the prefix and the category value. Default '_' gives city_Paris; use '=' for city=Paris.", "type": "string" }, "sort": { "default": "alphabetical", "description": "Order of the generated columns: 'alphabetical' by value, 'frequency' most common first, or 'first-seen' in the order values appear in the data. max_categories always selects by frequency regardless of this.", "enum": [ "alphabetical", "frequency", "first-seen" ], "type": "string" } }, "required": [ "data", "column" ], "type": "object" } } }