[Feature] Data exploration of unique rows

[jaychia]

Summary

An important part of data exploration is the retrieval and interrogation of uniqueness of values within column(s) as well as counts.

In SQL, this is achieved with:

1. SELECT DISTINCT x, y, z FROM ... - retrieves distinct tuples of (x, y, z)
2. SELECT x, y, z, COUNT(*) GROUP BY x, y, z - retrieves distinct tuples of (x, y, z) and counts of their occurrences
3. SELECT COUNT(DISTINCT x), COUNT(DISTINCT y), COUNT(DISTINCT z) - retrieves counts of distinct occurrences of x, y, and z

In other DataFrames such as Pandas, there are dedicated methods such as:

1. df.drop_duplicates
2. df.groupby(x, y, z).size()
3. df.value_counts

Proposal

# 1: Retrieve a DataFrame of distinct rows - equivalent to a SQL `SELECT DISTINCT` or `pd.drop_duplicates`
df = df.distinct(df["x"], df["y"], df["z"])

# 2: Count occurrences of distinct tuples
# Alias for: df.groupby(df["x"], df["y"], df["z"]).agg([("*", "count")])
df.groupby(df["x"], df["y"], df["z"]).count()

# 3: Retrieve counts of distinct occurrences
df.agg([
    (df["x"], "count_distinct"),
    (df["y"], "count_distinct"),
    (df["z"], "count_distinct"),
])

Work Breakdown

Concretely, the work needed to achieve the above is:

1. Allow .distinct() to take in columns as arguments
2. Implement aggregations on groupby keys
3. Implement a `"count_distinct" aggregation