The FI311 Unit 7 fairness memo here traces an age gap to device, phone and email features, then recommends dropping them for a less discriminatory model that costs little accuracy. Searches like "fi 311 unit 7 assignment example", "fi311 unit 7 sample" and "fi311 unit 7 example" land here.
What a finished FI311 Unit 7 algorithmic fairness memo looks like
Four pages in memo format, recommendation first. The opening paragraph asks the committee to approve removing three features and adopting an alternative model already tested. The findings section reports composite fair lending results: applicants 62 and older are approved at 0.79 times the rate for applicants aged 25 to 61, below the four-fifths ratio the lender uses as a screening threshold, and applicants from predominantly Hispanic ZIP codes, identified by a proxy method, at 0.83. A driver analysis traces most of the age gap to device age, landline-only phone numbers and email addresses issued by internet providers. The legal section states the equal credit rule narrowly and explains disparate impact. The alternative model raises the age ratio to 0.91 while losing 0.4 points of predictive accuracy.
How a FI311 Unit 7 example is structured
Request, evidence, law, options, decision. The memo's first paragraph carries the entire request, sized for a committee agenda. Findings follow as a small table of approval ratios by group, with a note on how group membership was estimated, since the lender does not collect race or ethnicity for these loans and relies on a surname and geography proxy. The driver analysis explains why an older phone, a landline number and an internet-provider email address correlate with age, and how much predictive value each adds. The legal section states the Equal Credit Opportunity Act narrowly: no discrimination on prohibited bases, age among them, and specific principal reasons owed to declined applicants. An options table compares keeping the model, dropping the three features, and adopting the tested alternative. The decision section costs each option.
A request the committee can act on
Remove three features, adopt the tested alternative, and retest quarterly: the ask appears in the first paragraph with its accuracy cost attached.
Approval ratios by group
The 0.79 age ratio and 0.83 ethnicity ratio are shown against the lender's four-fifths screen, with the proxy method's limits stated beside the second figure.
Three features that track age
Older handsets, landline-only numbers and internet-provider email domains are far commoner among older applicants and add little predictive value once income and payment history are in.
What the equal credit rule asks
Prohibited bases, including age, may not drive credit decisions, declined applicants are owed specific principal reasons, and the status of disparate impact enforcement is flagged as contested.
An alternative that costs 0.4 points
The retrained model without the three features lifts the age ratio to 0.91 and the ethnicity ratio to 0.90, at a predictive cost the memo prices in expected losses.
Where marks go in FI311 Unit 7
Declaring a model biased, or clean, without a measurement is the quickest way for an FI311 fairness memo to lose marks. A ratio by group, with the method for estimating group membership stated, gives the committee something to act on; an adjective does not. Memos that stop at removing a protected characteristic come next, as if a model that never sees age cannot sort by it. Proxies are the substance of this unit. Recommendations that ignore accuracy draw comment, because a lender will not adopt a fix whose cost is unstated. Legal claims need care: what the rule requires stated narrowly, and unsettled enforcement positions flagged as unsettled. Memos that also address the reasons owed to declined applicants, and whether a feature like handset age could ever be explained to one, show the fullest grasp.
Get a FI311 Unit 7 example written to your instructions
Share the model, the lender and any test results your Unit 7 prompt supplies, with the FI311 memo instructions and rubric. Who is disadvantaged gets measured first, then the responsible features are traced, the governing rule stated narrowly and each option costed. It remains a course memo rather than a fair lending review of any actual lender. First custom sample free, back within 24-48h.
FI311 Unit 7 questions, answered
How can a model discriminate by age if age is not an input?
Through proxies. Variables that look neutral, such as an older phone or a landline number, are far more common among older applicants, so a model that penalizes them penalizes age indirectly. The sample measures outcomes by age group rather than inspecting inputs, which is how the gap was found, and then works backward to the features producing it.
What is the four-fifths ratio, and is it a legal standard?
It compares the approval rate of one group with that of the most favored group, and a result below 0.80 is commonly treated as a signal worth investigating. It began as a rule of thumb in employment selection and is not a legal line in lending. The sample uses it as the lender's internal screening threshold and says so plainly.
Why does the memo flag disparate impact as contested?
Because the federal posture on enforcing it in lending has moved recently, and a memo that assumes either answer risks being wrong by the time it is read. The sample states the long-standing interpretation, notes that it has been challenged, and builds its recommendation on grounds that hold either way: the three features add little accuracy and cannot be explained to a declined applicant.