Reading Proficiency by School

State's description: accountability - indicator data - performance composite (EOG/EOC)

Listed in the state's table index as rcd_acc_pc (active, 2014)

972,809 rows loaded, covering 2,395 schools and agencies, 2013–14 to 2023–24.

1. The state publishes an Excel data dictionary. This demo reads it: 66 table definitions, 525 column definitions, and 181 code definitions — the same ones behind the Decoded view above.
2. From those definitions it writes a Django database model for each table, so nobody has to type the columns in by hand. This table's model came out like this:

class RcdAccPc(SRCBaseModel):
    agency_code = models.CharField(max_length=6)
    den = models.DecimalField(max_digits=38, decimal_places=0, null=True, blank=True)
    grade = models.CharField(
        max_length=16,
        choices=[
            ("PK", "Pre-kindergarten"),
            ("KG", "Kindergarten"),
            ("01", "Grade 01"),
            ("02", "Grade 02"),
            ("03", "Grade 03"),
            ("04", "Grade 04"),
            ("05", "Grade 05"),
            ("06", "Grade 06"),
            ("07", "Grade 07"),
            ("08", "Grade 08"),
            ("09", "Grade 09"),
            ("10", "Grade 10"),
            ("11", "Grade 11"),
            ("12", "Grade 12"),
            ("13", "Grade 13"),
        ],
    )
    masking = models.CharField(
        max_length=192,
        null=True,
        blank=True,
        choices=[
            ("1", ">95%"),
            ("2", "<5%"),
            ("3", "<10 in denominator"),
            (
                "4",
                "Insufficient Data (per business area, specific alternate minimum denominator, there is not enough"
                " data to provide a meaningful answer - additional set of masking rules on a case by case basis)",
            ),
        ],
    )
    pct = models.DecimalField(max_digits=4, decimal_places=1, null=True, blank=True)
    standard = models.CharField(max_length=3)
    subgroup = models.CharField(
        max_length=40,
        choices=[
            ("EDS", "Economically Disadvantaged"),
            ("NEDS", "Not Economically Disadvantaged"),
            ("NOT_EDS", "Not Economically Disadvantaged"),
            ("ELS", "English Learners"),
            ("NELS", "Not English Learners"),
            ("SWD", "Students With Disabilities"),
            ("NSWD", "Not Students With Disabilities"),
            ("WH7", "White"),
            ("BL7", "Black"),
            ("HI7", "Hispanic"),
            ("AM7", "American Indian"),
            ("AS7", "Asian"),
            ("MU7", "Two or More Races"),
            ("MALE", "Male"),
            ("FEM", "Female"),
            ("ALL", "All Students"),
            ("AIG", "Academically / Intellectually Gifted"),
            ("NAIG", "Not Academically / Intellectually Gifted"),
            ("PI7", "Pacific Islander"),
            ("HMS", "Homeless"),
            ("FCS", "Foster Care"),
            ("MIL", "Military Connected"),
            ("MIG", "Migrant"),
            ("AIAN", "American Indian / Alaskan Native"),
            ("ASPI", "Pacific Islander / Asian"),
            ("Other", "Other"),
        ],
    )
    subject = models.CharField(max_length=6)
    year = models.CharField(max_length=4)

    class Meta:
        db_table = "rcd_acc_pc"
        managed = True
        unique_together = ("year", "agency_code", "standard", "subject", "grade", "subgroup")

3. 972,809 rows were then loaded from the state's raw data files into RcdAccPc.
4. It also takes a fingerprint (SHA-256) of the dictionary's general-rules sheet. If the state quietly rewrites its rules document, the next rebuild flags the change instead of absorbing it silently.