Struct Summary 

pub struct Summary {

    pub trained_at: i64,
    pub seconds: f64,
    pub tokenizer: u32,
    pub features: u32,
    pub replaced: bool,
    pub reason: String,
    pub threshold_spam: f64,
    pub threshold_unsure: f64,
    pub labels: Summary,
    pub split: Split,
    pub accounts: BTreeMap<String, Counts>,
    pub model: ModelSummary,
    pub test: Numbers,
    pub test_settled: Option<Numbers>,
    pub current: Option<Numbers>,
    pub compared: Option<Compared>,
    pub refit: Option<Refit>,
    pub outside: BTreeMap<String, Outside>,
    pub trial: bool,
}

The last training, as trained.toml keeps it for the settings: numbers only.

Fields

trained_at: i64

When it ended (Unix seconds), and how long it took.

seconds: f64

tokenizer: u32

features: u32

replaced: bool

The table took this training’s model.

reason: String

Why: “no-table” (none yet), “unreadable” (the one in place was made by another version, or damaged), “nothing-new” (no test message came after the one in place was trained), “no-worse”, “worse” (kept).

threshold_spam: f64

threshold_unsure: f64

labels: Summary

split: Split

accounts: BTreeMap<String, Counts>

Per account: ham and spam learned from, ham and spam tested on.

model: ModelSummary

test: Numbers

The numbers, on the newest 20 %, of the model learned from the oldest 80 % (model): what decided the replacement.

test_settled: Option<Numbers>

The same, ham an inbox holds since less than detail::UNSETTLED_DAYS left out: its label may not be settled (spam nobody looked at yet).

current: Option<Numbers>

The table in place before, on the test messages it never learned from (those after its training: compared).

compared: Option<Compared>

What the replacement compared.

refit: Option<Refit>

The table written, when replaced: the same model learned again from every message.

outside: BTreeMap<String, Outside>

Outside material learned from, by source (external): what each gave, and its baseline.

trial: bool

A trial (Options::replace false): nothing was replaced nor written; reason says what the comparison found (“worse”: the table in place would have stayed).

Trait Implementations

impl Clone for Summary

fn clone(&self) -> Summary

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Summary

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for Summary

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl PartialEq for Summary

fn eq(&self, other: &Summary) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl Serialize for Summary

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.

impl StructuralPartialEq for Summary