Count by category and track the largest frequency

A record’s category may repeat much later in a shuffled pile. Keep one persistent count for each category, rather than resetting whenever the next card has a different label.

Create a counter when the category first appears #

If you have not seen the category, create its counter at zero. Then add 1 for the current record. The creation and increment are separate ideas: the first record of a category must not be missed.

Counts = empty table
MaxCount = 0
For each record X {
    key = X.category
    If key has no counter: Counts[key] = 0
    Counts[key] = Counts[key] + 1
    if (Counts[key] > MaxCount) {
        MaxCount = Counts[key]
    }
}

Counts[key] means “the counter belonging to this category” in a small labeled table. It is teaching notation for the lecture’s one-variable-per-category idea; you do not need to know a specific programming-language dictionary yet.

Why compare only the updated counter? #

Every other counter is unchanged. If MaxCount was correct before this visit, only the changed counter can beat it. Counts grow by one, so the previous best value never becomes invalid by decreasing. This avoids a separate scan of all category counters after each record.

Lecture and tutorial results #

A bill count measures purchases represented by bills, not all visitors to a store and not necessarily distinct customers. A person can generate several bills. Also, keeping only MaxCount gives the winning number, not automatically the winning category; store its label if the question asks for it.

Source: Lecture 2.4; Tutorial 2.3. Small teaching datasets and diagrams are prepared for these notes; lecture and tutorial totals are identified explicitly.