This page collects 120 targeted questions and answers distilled from the Python implementation under python/PruebaTecnicaXpertgroupIngAWS26Dic2025Python to help you review behavior before a technical interview.
Thank you so much for accepting my connection request. I'm truly grateful to connect with you.
I also want to sincerely thank you for your role as an HR recruiter. Through your work, many are given the opportunity to receive their daily bread - a blessing that reflects the words of our Lord: "Give us this day our daily bread" (Matthew 6:11). Your dedication is part of how God's providence reaches others.
I also want to share that I read the following document every single day: PETICIONES-SANTO-ROSARIO and PETITIONS-HOLY-ROSARY.
It contains 33 Petitions to the Holy Rosary, and in it, I offer blessings and prayers for all the people to whom I send good messages - and today, that includes you.
May the Lord bless you abundantly, protect you, and guide you always.
Warm regards,
Manuela
PatientRecord capture and how does it infer categoria?The dataclass in src/core/models.py stores id_paciente, nombre, fecha_nacimiento, edad, sexo, email, telefono, ciudad, and categoria; PatientRecord.from_dict calls categorize_patient_age to compute categoria from the provided age or birthdate.
PatientCategory expose and when does code pick UNKNOWN?The enum lists CHILD, ADULT, SENIOR, and UNKNOWN; categorize_patient_age returns UNKNOWN when neither an integer age nor a parsable fecha_nacimiento is present.
categorize_patient_age apply when given an age versus a birthdate?If edad is an int it delegates to categorize_by_value; otherwise it tries to parse ISO-formatted fecha_nacimiento, calculates the difference versus the current UTC year, and defaults to UNKNOWN when parsing fails.
categorize_by_value bucket ages?Values below 18 map to CHILD, values below 65 map to ADULT, and all others become SENIOR.
CompletenessMetric track for each field?It records the field, the total and missing counters, the completeness ratio, and dictionaries per_city_missing and per_category_missing so stakeholders can see percentages per city and per patient category.
ImputationPlan describe?The dataclass bundles a target field, the chosen strategy, and a rationale string describing the recommended fix for missing data.
AgeCorrectionLogEntry record for inconsistencies?Each entry logs id_paciente, nombre, fecha_nacimiento, the registered and calculated ages, an action such as inconsistent_age or imputed_age, and a descriptive note.
AppointmentRecord.from_dict normalize the ciudad value?It prefers ciudad and falls back to ciudad_cita so each appointment retains a city even when only the appointment location is supplied.
AppointmentIndicatorEntry expose?The entry captures period_type, period_value, especialidad, estado_cita, medico, and the aggregated count.
CostAuditReport?The report stores total_records, analyzed_records, a list of summaries (SpecialtyCostSummary) with averages and deviations, and a list of anomalies (CostAnomalyEntry).
PatientTravelEntry log for each traveler?It records id_paciente, nombre, residence, the travel_cities set, travel_count, the computed severity, and last_travel_dates.
A BusinessRule provides an id, title, description, and flexible details; BusinessRulesCatalog aggregates those rules with a created_at timestamp.
PatientRepository and AppointmentRepository export?Each port defines a single abstract method, list_patients or list_appointments, so the core services depend on interfaces instead of concrete adapters.
They open the dataset path, parse JSON, fetch the pacientes or citas_medicas array, and yield domain objects through PatientRecord.from_dict or AppointmentRecord.from_dict.
backend/app.py point to by default?The module sets DATASET to BASE_PATH / "dataset_hospital 2 AWS.json", so every service call uses that JSON file unless it is replaced.
ETLPipelineService.run() perform?It records a start timestamp, calls extract, transform, and load, enriches the summary with patient and appointment counts plus start/end/duration, persists metrics, and returns that summary dictionary.
transform handle orphan appointments?After cleaning both dataframes, it applies a mask to drop rows whose id_paciente is not in the cleaned patient frame and notes the dropped id_citas as orphans.
The load step writes pacientes_cleaned.csv/.parquet and citas_cleaned.csv/.parquet under reports/etl.
The helper catches ImportError from df.to_parquet and writes an empty UTF-8 placeholder so the CSV still documents the table.
_persist_metrics keep track of ETL runs?It appends an entry to reports/etl/etl_metrics.json with start/end times, duration, exported counts, and orphan total, creating the file when it does not exist.
It computes BASE_PATH as two levels above app.py and defines REPORT_DIR and SCRIPTS_DIR relative to that root.
The loader looks for scripts/run_*.py, reads each docstring, normalizes the key, and registers descriptors used by the /scripts endpoints.
AccessibilityService.evaluate()?It cross compares appointment counts per patient with their residence city to surface patients whose volume significantly exceeds local peers.
The service builds overall and city-specific counts, computes mean and pstdev, and flags counts that exceed the city mean plus twice the city standard deviation or the global average plus global standard deviation when city data is sparse.
AccessibilityReport publish?It emits total_pacientes, flagged, and the entries list of AccessibilityEntry objects sorted by travel volume.
When their appointment count is above the city threshold and the city has enough data, or when the overall standard deviation is positive and their count exceeds overall average plus standard deviation.
AgeConsistencyService.audit_ages() report?It builds an AgeConsistencyReport with the cutoff date, total records, inconsistency count, imputations, missing birthdates, and log entries for each anomaly.
Entries with blank fecha_nacimiento get action missing_birthdate, while parsable failures get invalid_birthdate along with notes explaining why.
edad but has a valid birthdate?The service creates an AgeCorrectionLogEntry with action imputed_age and records the calculated value based on the cutoff date.
_calculate_age treat birthdays before the cutoff date?It subtracts birth years, decrements by one if the cutoff month/day precedes the birthday, and never returns a negative value.
AgeSpecialtyMismatchService?The service defines ranges for keywords containing pediatr (0 to 17) and geriatr (65 to 150).
It falls back to the default range of 18 to 64.
It returns None if either fecha_nacimiento or fecha_cita is missing or malformed, using safe ISO parsing.
AgeSpecialtyMismatchEntry?When the calculated age sits below the expected minimum or above the expected maximum for the normalized specialty.
AppointmentAlertEntry?The service flags appointments that lack a fecha_cita or that do not have an assigned medico.
It collects missing pieces into note_parts, joins them with a conjunction, and prefixes the phrase so reviewers know which fields to validate.
AppointmentAlertReport include?Total input records, the number of alerts, and the detailed entries list.
Because the loop continues whenever both falta_fecha and falta_medico are false, avoiding false positives.
AppointmentCostAuditService.analyze()?Costs are grouped by normalized specialty, then mean and standard deviation are computed per group while also keeping a map of processed records.
A record is anomalous when its cost deviates from the specialty average by more than twice the specialty standard deviation computed without the current appointment.
SpecialtyCostSummary entries expose?The properties expected_min and expected_max subtract or add twice the standard deviation from the average.
_normalize_specialty provide for blank values?It returns the trimmed specialty string when present or sin_especialidad when the field is empty.
measure_indicators capture?The method tallies counts for daily buckets and weekly buckets (ISO week), producing entries for both.
Each appointment without a parseable fecha_cita increments missing_dates.
AppointmentIndicatorReport contain?It reports total_records, missing_date, the combined entries, and the top five bottlenecks.
_safe sanitize values?It strips strings, ensures they are not empty, and falls back to defaults such as sin_especialidad or sin_medico.
Only records whose estado_cita is Completada or Cancelada are processed.
_collect_issues detect?It notes absent fecha_cita or missing medico, adding descriptive strings for the report.
AppointmentReviewReport?The report returns total_citas, reviewed_citas, and the entries list of problematic appointments.
The service focuses on completed and canceled visits for quality control, so it skips states outside the target set.
AppointmentStateTimelineService reconstruct state history?It groups records by id_cita, sorts them by parsed dates, and builds ordered transitions lists.
Reprogramming events increment counters per doctor-week and track affected appointment IDs inside the occupancy dict.
Any status whose lowercase string starts with reprogram increments the reprogram counter and contributes to occupancy data.
It reads the last transition value (or defaults to sin_estado) after sorting all recorded states.
It describes valid appointment states, age ranges per specialty, email format, and phone format along with identifiers such as rule-estado and rule-edad-especialidad.
The service uses datetime.utcnow().isoformat() for the catalog created_at field.
The email rule stores the regex ^[^@\s]+@[^@\s]+\.[^@\s]+$ inside its details dictionary.
They can read the documented states, ranges, and format expectations to align validation or governance checks with the expected contract.
The service uses _risk_threshold equal to 0.6 so scores at or above that value are considered high risk.
Previous cancellations or reprograms, short gaps between visits, current cancellation status, and the specialty weight from _specialty_weights each add to the weight while logging explanatory factors.
The helper _sigmoid applies 1 / (1 + exp(-weight)) so the final risk_score remains between zero and one.
_compute_days_between avoid negative values?It parses the current ISO date, subtracts the previous datetime, and returns the non negative day difference with max(delta.days, 0).
register_changes expect per event?Each event provides table, field, action, user, and optionally a timestamp and note.
The constructor indexes the provided FieldResponsibility objects so register_changes can look up the owner/contact pair for each table-field key.
The code defaults to datetime.utcnow().isoformat() so every entry remains timestamped.
CleaningAuditReport wrap?It returns a report with generated_at and the list of CleaningAuditEntry objects collected from the change events.
CompletenessService evaluate?The service checks email, telefono, and ciudad.
Counters accumulate total and missing values per city and patient category, then the metric converts them into percentages inside per_city_missing and per_category_missing.
ImputationStrategy play?The service delegates to the supplied imputer with suggest(records), letting custom implementations return ImputationPlan recommendations.
evaluate return both metrics and the imputation plan?The tuple gives observers a snapshot of current data quality and a prioritized list of fields to fix.
DemandForecastService project demand?The constant MONTHS_AHEAD equals 3 so it builds projections for the next three months beyond the last observed month.
For each doctor, the service multiplies the last seen monthly count by 1.2, rounds to an integer, and trusts BASE_CAPACITY of 15 as the minimum.
_compute_avg_growth treat zero periods?It skips any month pair where the prior count is zero and returns zero if no valid growth rates exist.
DemandForecastReport provide?Besides entries it includes generated_at, avg_monthly_growth, months_ahead, future_months, and total_capacity.
The service looks for repeated appointments close in time (within seven days) or recurrent cancellations to build patterns.
Entries with more than one detected pattern become high severity, while single-pattern alerts remain medium.
patterns?The helper returns phrases such as "Multiples citas en periodos cortos" and "Cancelaciones recurrentes".
It sorts the entries by the number of patterns in descending order before building DoctorNotificationReport.
DoctorUtilizationService?UTILIZATION_THRESHOLD is 0.75 and CANCELLATION_THRESHOLD is 0.2 to decide when utilization is low or cancellations are high.
It divides completed appointments by the total scheduled count per doctor-specialty pair.
Any doctor-specialty whose utilization falls below 0.75 or whose cancellation rate (cancels plus reprograms over scheduled) exceeds 0.2 gets appended.
deviation field measure?It stores utilization - UTILIZATION_THRESHOLD so readers can see how far the rate is from the target.
DuplicateDetectionService apply?It groups patients by normalized name plus birthdate and by normalized name plus city to detect potential duplicates.
It sorts each group by the completeness score (count of filled attributes) and keeps the record with the highest score and lowest id.
DuplicateConsolidationReport expose?The report returns total_records, total_groups, total_duplicates, and the log_entries.
_completeness_score work?It awards one point for each non empty value among fecha_nacimiento, edad, sexo, email, telefono, and ciudad.
The definition list references referential_integrity_log.json, appointment_alerts_log.json, appointment_review_log.json, appointment_cost_audit_log.json, and reports/appointment_state_timeline_log.json.
If _read_value returns one or more items the category is included with its count.
The service sets CHANNEL to gobernanza@hospital.local.
_read_value navigate nested structures?It splits the path_expr on dots, iterates through dicts or digit indexes for lists, and returns a numeric count or zero when navigation fails.
ManagementKpiService produce?It returns the overall average cost and wait days plus ManagementKpiEntry rows per specialty with counts, average costs, and average waits.
The service sorts each patient timeline by appointment date and subtracts consecutive dates in days to accumulate global_waits and specialty buckets.
_normalize_specialty sanitize labels?It strips whitespace, title cases the value, and falls back to General when empty.
Costs are added only when estado_cita trimmed and lowercased equals completada and the costo field is not null.
The service maps COMPLETED, CANCELED, and REPROGRAMMED label sets to normalized buckets before counting.
It uses a map from patient id to normalized city and pairs it with the normalized specialty before incrementing the dictionary bucket.
CitySpecialtyOccupancy entry contain?It tracks the city, specialty, and counts for completed, canceled, and reprogrammed appointments that share the same buckets.
total_appointments computed?The code increments the counter for each appointment whose status can be normalized into the predefined categories.
PatientSegmentationService?It uses ranges 0-17 (kids), 18-34 (young adults), 35-64 (adults), and 65+ (seniors).
Zero appointments become "Sin citas registradas", one to two appointments become "Frecuencia baja", three to four as "Frecuencia moderada", and five or more as "Frecuencia alta".
PatientSegmentationReport include?It reports generated_at, the total_patients, and the sorted list of PatientSegment cohorts.
_normalize_sex label gender?It maps variants of female to Femenino, male to Masculino, returns No declarado when missing, and title cases other values.
PatientTravelService flag?Patients whose appointment city differs from their residence are aggregated, and travelers must record more than one distinct city to be flagged.
A severity of high requires more than one visited city, otherwise the severity is medium.
PatientTravelReport provide?It lists generated_at, total_travelers, and the sorted travel entries.
ReferentialIntegrityService capture?It logs appointments with missing id_paciente or whose id_paciente is absent from the patient repository.
ReferentialIntegrityEntry store?Each entry holds id_cita, the optional id_paciente, and a motivo string describing the problem.
ReferentialIntegrityReport provide?It returns total_citas, orphan_citas, and the list of entries.
TextNormalizationService apply?It strips whitespace, lowercases, and removes diacritics via unicodedata normalization to deliver the NORMALIZATION_METHOD.
It iterates over nombre and ciudad.
TextNormalizationReport include?It returns total_records, a normalized_fields count, and the normalization log_entries.
QualityKpiService compare before and after cleaning?It evaluates the pacientes table and the citas_medicas table, computing metrics for both the before and after repositories.
FieldQualityMetric?Each metric reports completeness, uniqueness, and format_valid for the field.
The PATIENT_FIELDS list uses _is_valid_date and _is_valid_email where relevant, while APPOINTMENT_FIELDS requests _is_valid_date only for fecha_cita.
It lists all JSON report filenames under reports by reading REPORT_DIR.glob("*.json").
GET /reports/{name}/html returns a FileResponse streaming the corresponding HTML file after checking the path exists.
The endpoint calls _service_runner with the requested case, runs the selected service (doctor notifications, utilization, patient travel, management KPIs, or ETL), and returns the summary dictionary or report.to_dict().
It returns ScriptInfo objects for each discovered script, including the normalized key, module path, description, and relative path.
It ensures the uploaded file has content type application/json and is not empty before overwriting the dataset file.
A JSON payload with status ok and the dataset path string so callers can confirm the API and dataset location.