Files
integreat/scratch-sessions/ingest_ml.clj

72 lines
2.4 KiB
Clojure

(ns ingest-ml
(:require [datomic.client.api :as dc]
[clojure.java.io :as io]
[clojure.data.csv :as csv]
[iol-ion.tx :refer [upsert-entity]]
[auto-ap.datomic :refer [conn]]))
(println "hi")
(defn boot []
(user/init-repl)
(user/start-db))
(defn reset-inference []
(doseq [p (->>
(dc/q '[:find ?t
:where [?t :transaction/recommended-account]]
(dc/db conn))
(map (fn [[t]]
`(upsert-entity
~{:db/id t
:transaction/recommended-account nil
:tranasction/vendor-confidence nil
:transaction/account-confidence nil
:transaction/recommended-vendor nil})))
(partition-all 100))]
(dc/transact conn {:tx-data p})))
(defn read-inference []
(with-open [reader (io/reader "/mnt/data/dev2/ml-test/inference-outcome.csv")]
(->> (csv/read-csv reader)
(into []
(comp
(drop 1)
(map (fn [[_ transaction best-vendor best-account account-confidence]]
{:db/id (Long/parseLong transaction)
:transaction/recommended-account (Long/parseLong best-account)
:transaction/account-confidence (Double/parseDouble account-confidence)
:transaction/recommended-vendor (Long/parseLong best-vendor)}
)))))))
(defn apply-inference [inference]
(doseq [p (->> inference (partition-all 100))]
(dc/transact conn {:tx-data p})))
(defn check-applied-inference []
(clojure.pprint/pprint
(->>
(dc/q '[:find ?code ?bac ?do ?amount ?vc ?rvn ?ac ?ran
:in $
:where
[?t :transaction/recommended-account ?ra]
[?ra :account/name ?ran]
[?t :transaction/account-confidence ?ac]
[?t :transaction/recommended-vendor ?rv]
[?rv :vendor/name ?rvn]
[?t :transaction/description-original ?do]
[?t :transaction/client ?c]
[?c :client/code ?code]
[?t :transaction/bank-account ?ba]
[?ba :bank-account/code ?bac]
[?t :transaction/amount ?amount]
]
(dc/db conn))
(shuffle)
(take 10))))