(cloud) adds transaction insights

This commit is contained in:
2023-04-18 11:32:58 -07:00
parent ed37b19d6a
commit 69eaf42eda
11 changed files with 518 additions and 3 deletions

View File

@@ -0,0 +1,172 @@
;; This buffer is for Clojure experiments and evaluation.
;; Press C-j to evaluate the last expression.
;; You can also press C-u C-j to evaluate the expression and pretty-print its result.
(ns build-ml
(:require [datomic.client.api :as dc]
[clojure.java.io :as io]
[clojure.data.csv :as csv]
[auto-ap.datomic :refer [conn]]))
(println "hi")
(defn boot []
(user/init-repl)
(user/start-db))
(defn write-vendor-training-data []
(with-open [f (io/writer "/mnt/data/dev2/ml-test/input/vendor-training.csv")]
(csv/write-csv f
(into [["transaction" "client" "bank" "bank_type" "description" "date" "amount" "vendor"]]
(->>
(dc/q '[:find ?t ?c ?bn ?bt2 ?do ?d ?amt ?v
:in $ ?start
:where
[?t :transaction/date ?d]
[(>= ?d ?start)]
[?t :transaction/approval-status :transaction-approval-status/approved]
(not [?t :transaction/payment])
[?t :transaction/client ?c]
[?t :transaction/vendor ?v]
[?v :vendor/name ?v-name]
[?t :transaction/description-original ?do]
[?t :transaction/amount ?amt]
[?t :transaction/bank-account ?b]
[?b :bank-account/name ?bn]
[?b :bank-account/type ?bat]
[?bat :db/ident ?bt]
[(name ?bt) ?bt2]
]
(dc/db auto-ap.datomic/conn)
#inst "2021-01-01")
(map (fn [[t code bn bat d date amt v]]
[t code bn bat d (auto-ap.time/unparse-local (clj-time.coerce/to-date-time date) auto-ap.time/iso-date) amt v]))))
:quote? (constantly true))))
(defn write-account-training-data []
(with-open [f (io/writer "/mnt/data/dev2/ml-test/input/account-training.csv")]
(csv/write-csv f
(into [["transaction" "client" "bank" "bank_type" "description" "date" "amount" "vendor" "account"]]
(->>
(dc/q '[:find ?t ?c ?bn ?bt2 ?do ?d ?amt ?v ?a
:in $ ?start
:where
[?t :transaction/date ?d]
[(>= ?d ?start)]
[?t :transaction/approval-status :transaction-approval-status/approved]
[?t :transaction/accounts ?ta]
[?ta :transaction-account/account ?a]
[?t :transaction/vendor ?v]
[?v :vendor/name ?v-name]
[?t :transaction/client ?c]
[?t :transaction/description-original ?do]
[?t :transaction/amount ?amt]
[?t :transaction/bank-account ?b]
[?b :bank-account/name ?bn]
[?b :bank-account/type ?bat]
[?bat :db/ident ?bt]
[(name ?bt) ?bt2]]
(dc/db auto-ap.datomic/conn)
#inst "2022-01-01")
(map (fn [[t code bn bat d date amt v a]]
[t code bn bat d (auto-ap.time/unparse-local (clj-time.coerce/to-date-time date) auto-ap.time/iso-date) amt v a]))))
:quote? (constantly true))))
(defn write-inference []
(with-open [f (io/writer "/mnt/data/dev2/ml-test/input/inference.csv")]
(csv/write-csv f
(into [["transaction" "client" "bank" "bank_type" "description" "date" "amount"]]
(->>
(dc/q '[:find ?t ?c ?bn ?bt2 ?do ?d ?amt
:in $ ?start
:where
[?t :transaction/date ?d]
[(>= ?d ?start)]
[?t :transaction/approval-status :transaction-approval-status/unapproved]
(not [?t :transaction/matched-rule])
(not [?t :transaction/payment])
[?t :transaction/client ?c]
[?c :client/code ?code]
(not [?t :transaction/vendor])
(not [?t :transaction/accounts])
[?t :transaction/description-original ?do]
[?t :transaction/amount ?amt]
[?t :transaction/bank-account ?b]
[?b :bank-account/name ?bn]
[?b :bank-account/type ?bat]
[?bat :db/ident ?bt]
[(name ?bt) ?bt2]
]
(dc/db auto-ap.datomic/conn)
#inst "2023-01-01")
(map (fn [[t code bn bat d date amt]]
[t code bn bat d (auto-ap.time/unparse-local (clj-time.coerce/to-date-time date) auto-ap.time/iso-date) amt]))
))
:quote? (constantly true))))
(defn write-vendor-lookup []
(with-open [f (io/writer "/mnt/data/dev2/ml-test/input/vendors.csv")]
(csv/write-csv f
(into [["vendor" "vendor_name"]]
(->>
(dc/q '[:find ?v ?v-name
:where [?v :vendor/name ?v-name]
[_ :transaction/vendor ?v]
]
(dc/db conn))))
:separator \,)))
(defn write-vendor-client-lookup []
(with-open [f (io/writer "/mnt/data/dev2/ml-test/input/vendor-client-usage.csv")]
(csv/write-csv f
(into [["vendor" "client" "vendor_client_count"]]
(->>
(dc/q '[:find ?v ?c (count ?t)
:where
[?t :transaction/client ?c]
[?t :transaction/vendor ?v]
]
(dc/db conn))))
:separator \,)))
(defn write-account-lookup []
(with-open [f (io/writer "/mnt/data/dev2/ml-test/input/accounts.csv")]
(csv/write-csv f
(into [["account" "numeric_code"]]
(->>
(dc/q '[:find ?a ?nm
:where [?a :account/numeric-code ?nm]]
(dc/db conn))))
:separator \,)))
(defn write-account-client-lookup []
(with-open [f (io/writer "/mnt/data/dev2/ml-test/input/account-client-usage.csv")]
(csv/write-csv f
(into [["client" "account" "client_account_count"]]
(->>
(dc/q '[:find ?c ?a (count ?t)
:where
[?t :transaction/client ?c]
[?t :transaction/accounts ?ta]
[?ta :transaction-account/account ?a]]
(dc/db conn))))
:separator \,)))
(defn write-all-training []
(write-account-lookup)
(write-vendor-lookup)
(write-account-client-lookup)
(write-vendor-client-lookup)
(write-account-training-data)
(write-vendor-training-data))

View File

@@ -0,0 +1,71 @@
(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))))