# Chapter 1
```elixir
Mix.install([
{:axon, "~> 0.7"},
{:nx, "~> 0.9"},
{:explorer, "~> 0.10.0"},
{:kino, "~> 0.14.2"}
])
```
## Section
```elixir
require Explorer.DataFrame, as: DF
iris = Explorer.Datasets.iris()
```
```elixir
cols = ~w(sepal_width sepal_length petal_length petal_width)
normalized_iris =
DF.mutate(
iris,
for col <- across(^cols) do
# Needed to cast result to :f32 due to
{col.name, cast((col - mean(col)) / standard_deviation(col), :f32)}
end
)
```
```elixir
normalized_iris = DF.mutate(normalized_iris, [
species: Explorer.Series.cast(species, :category)
])
```
```elixir
shuffled_normalized_iris = DF.shuffle(normalized_iris)
```
```elixir
train_df = DF.slice(shuffled_normalized_iris, 0..119)
test_df = DF.slice(shuffled_normalized_iris, 120..149)
```
```elixir
feature_columns = [
"sepal_length",
"sepal_width",
"petal_length",
"petal_width"
]
x_train = Nx.stack(train_df[feature_columns], axis: -1)
y_train =
train_df["species"]
|> Nx.stack(axis: -1)
|> Nx.equal(Nx.iota({1, 3}, axis: -1))
x_test = Nx.stack(test_df[feature_columns], axis: -1)
y_test =
test_df["species"]
|> Nx.stack(axis: -1)
|> Nx.equal(Nx.iota({1, 3}, axis: -1))
```
```elixir
model =
Axon.input("iris_features", shape: {nil, 4})
|> Axon.dense(3, activation: :softmax)
```
```elixir
Axon.Display.as_graph(model, Nx.template({1, 4}, :f64))
```
```elixir
data_stream =
Stream.repeatedly(fn ->
{x_train, y_train}
end)
```
```elixir
trained_model_state =
model
|> Axon.Loop.trainer(:categorical_cross_entropy, :sgd)
|> Axon.Loop.metric(:accuracy)
|> Axon.Loop.run(data_stream, Axon.ModelState.new(%{}), iterations: 500, epochs: 10)
```
```elixir
data = [{x_test, y_test}]
model
|> Axon.Loop.evaluator()
|> Axon.Loop.metric(:accuracy)
|> Axon.Loop.run(data, trained_model_state)
```
