Lesson 1Building of buy tables: buys, buy parts, returns, life value signs and field picksLearn to build main buy tables that catch buys, line parts, returns, and life value signs. We talk key fields, normal choices, and how to help later checks and suggestion works.
Buy head vs line part plan buildModelling returns, money-backs, and stopsCatching cuts, tickets, and feesStoring life value and edge signsKeys, lists, and part choicesLesson 2Dealing with rough and thin act data: meeting-making, bot clean, same-remove, happening weightLook at ways to clean rough act logs and make thin data useful. You will learn meeting rules, bot and grabber clean, same-remove reason, and happening weight plans fit to suggestion train.
Meeting rules and time-outsFinding and clean bots and grabbersClick, see, and buy same-removeHappening weight for model trainHandling thin users and new startsLesson 3Building of goods list table: goods mark, title, kind tree, parts, price, mark, store, pictures, main text, embeddingsLearn to shape a goods list table that helps fast get and rich suggestions. We cover marks, parts, prices, store, media, main text, and embeddings, plus ways for updates and un-normal.
Steady goods and change marksKind tree and partsPrice, store, and ready fieldsPictures, media, and main textStoring and update item embeddingsLesson 4Part making rules for suggestions: newness, times, money, item like, kind like, user embeddingsFind main part making rules for suggester systems. We detail newness, times, money value, like, kind like, and user embeddings, including group times and leak-safe count ways.
Newness, times, and money partsItem and kind like signsUser-kind and mark like scoresOrder-based and meeting partsUser and item embedding makeLesson 5Extra data sets: item info, kind list, boosts, content (words), seller dataUnderstand how extra data sets make suggestions richer than raw clicks and buys. We cover item info, kind list, boosts, content, and seller feeds, plus how to keep them steady, versioned, and joinable at big scale.
Building item info plansKeeping goods kind treesModelling boosts and price rulesStoring rich content and wordsJoining seller and feed dataLesson 6Data clean and fill ways: gone parts, price odd, wrong timesLearn real data clean and fill methods for e-commerce. We fix gone parts, odd prices, wrong times, and uneven moneys, focusing on rules, short ways, and effect on suggestion strength.
Finding and fix gone partsHandling out and zero pricesFixing wrong or rough timesMoney, fee, and unit normalWriting clean rules and effectsLesson 7Building of happening flow and act table: happening mark, user mark/meeting mark, happening kind, goods mark, time, place (sender, page kind), device, happening valueBuild a joined act table and happening flow that catches user acts across paths. Learn happening plans, marks, place fields, and how to help both real-time flow and off-line batch suggestion lines.
Picking happening and user marksModelling happening kinds and partsCatching place, device, and senderHappening time, take time, and orderFlow vs batch store waysLesson 8Building of user files table: need fields (user mark, join time, email hash, people info, life step, groups, opt-in flags) and reasonBuild a user files table that balances personal power with secret and rule keep. We cover need fields, life and groups, opt-in flags, hash secret data, and how files feed suggestion models.
Main marks and join infoPeople info and life stepsAct and selling groupsAgree, opt-in, and like flagsSecret, hash, and keep rules
