libraryDependencies += "io.github.edadma" %%% "tabula" % "0.0.1"

import io.github.edadma.tabula.*

// From CSV file
val iris = Dataset.fromCSVFile("iris.csv")

// From in-memory data
val sales = Dataset(
  Seq("product", "revenue", "units"),
  Seq(
    Seq("Widget A", 1250.50, 45),
    Seq("Widget B", 890.25, 32),
    Seq("Widget C", 2100.75, 67)
  )
)

// From a Map
val temperatures = Dataset(Map(
  "city" -> Seq("New York", "London", "Tokyo"),
  "temp_c" -> Seq(22.5, 18.2, 25.1),
  "humidity" -> Seq(65, 78, 82)
))

// Display dataset info
iris.info()
// <class io.github.edadma.tabula.Dataset>
// 150 rows; 5 columns
// #  Column        Non-Null Count  Datatype
// 0  sepal_length  150             float
// 1  sepal_width   150             float
// 2  petal_length  150             float
// 3  petal_width   150             float
// 4  species       150             string

// View first/last rows
iris.head()     // First 5 rows
iris.tail(10)   // Last 10 rows

// Get descriptive statistics
iris.describe.print()
//       count  mean   std    min    25%    50%    75%    max
// sepal_length  150.0  5.8433  0.8281  4.3000  5.1000  5.8000  6.4000  7.9000
// sepal_width   150.0  3.0573  0.4359  2.0000  2.8000  3.0000  3.3000  4.4000
// ...

// Access columns
val sepalLength = iris("sepal_length")
val species = iris.species  // Dynamic column access

// Filter by conditions
val largeSepals = iris(iris("sepal_length") > 6.0)
val setosa = iris(iris("species") == "setosa")

// Complex filtering
val filtered = iris(
  (iris("sepal_length") > 5.0) && 
  (iris("petal_width") < 2.0)
)

// Select specific rows by index
val firstTen = iris.iloc(0 to 9)
val sample = iris.sample(20)  // Random sample of 20 rows

// Column statistics
iris("sepal_length").mean    // Mean value
iris("sepal_length").std     // Standard deviation
iris("sepal_length").min     // Minimum value
iris("sepal_length").max     // Maximum value

// Z-score normalization
val normalized = iris.zcode

// Remove outliers (keep values within 3 standard deviations)
val cleaned = iris((iris.zcode.abs < 3).all)

// Sort by column
val sorted = iris.sort("sepal_length")

// Add new columns
val enhanced = iris.append(
  (iris("sepal_length") * iris("sepal_width")).rename("sepal_area")
)

// Drop columns
val reduced = iris.drop("sepal_width", "petal_width")

// Sample and split data
val shuffled = iris.shuffle
val Array(train, test) = iris.split(80, 20)  // 80% train, 20% test

// Load timestamp data
val timeseries = Dataset(
  Seq("timestamp", "value"),
  Seq(
    Seq("2023-01-01 09:00:00", 100.5),
    Seq("2023-01-01 10:00:00", 102.3),
    Seq("2023-01-01 11:00:00", 98.7)
  )
)

// Time-based operations
val recent = timeseries(timeseries("timestamp") > Instant.parse("2023-01-01T09:30:00Z"))

import io.github.edadma.tabula.PG

val connection = PG.connect("localhost", "mydb", "user", "password", 5432)
val results = PG.query(
  "SELECT * FROM sales WHERE revenue > 1000", 
  connection
)

// Correlation and regression metrics
val predictions = model.predict(testData)
val error = testData.error("predictions", "actual")
val r2Score = testData.r2("predictions", "actual")
val adjustedR2 = testData.adjustedR2("predictions", "actual")

// Value counts and frequency analysis
val speciesCounts = iris.counts("species")
val normalized = iris.countsNormalize("species")

import io.github.edadma.tabula.Type.*

val dataset = Dataset(
  Seq("id", "score", "active", "created"),
  data,
  types = Seq(IntType, FloatType, BoolType, TimestampType)
)