RESUMO
Educators seek to harness knowledge from educational corpora to improve student performance outcomes. Although prior studies have compared the efficacy of data mining methods (DMMs) in pipelines for forecasting student success, less work has focused on identifying a set of relevant features prior to model development and quantifying the stability of feature selection techniques. Pinpointing a subset of pertinent features can (1) reduce the number of variables that need to be managed by stakeholders, (2) make "black-box" algorithms more interpretable, and (3) provide greater guidance for faculty to implement targeted interventions. To that end, we introduce a methodology integrating feature selection with cross-validation and rank each feature on subsets of the training corpus. This modified pipeline was applied to forecast the performance of 3225 students in a baccalaureate science course using a set of 57 features, four DMMs, and four filter feature selection techniques. Correlation Attribute Evaluation (CAE) and Fisher's Scoring Algorithm (FSA) achieved significantly higher Area Under the Curve (AUC) values for logistic regression (LR) and elastic net regression (GLMNET), compared to when this pipeline step was omitted. Relief Attribute Evaluation (RAE) was highly unstable and produced models with the poorest prediction performance. Borda's method identified grade point average, number of credits taken, and performance on concept inventory assessments as the primary factors impacting predictions of student performance. We discuss the benefits of this approach when developing data pipelines for predictive modeling in undergraduate settings that are more interpretable and actionable for faculty and stakeholders. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s41239-021-00279-6.
RESUMO
Clinical outcome of cocaine body packers is considered to be unpredictable and there are no clear guidelines for the management of these patients. Their surveillance in casualty wards, where they are usually admitted during evacuation of the packets, can be very difficult. The authors refer to a new type of cocaine packet, allowing these patients to be managed with a more conservative approach than in the past, and report their experience with 161 body packers, observed from January 1999 to December 2000. They adopted a surveillance protocol providing only minimal medical intervention. Among 161 body packers, 142 (88.2%) evacuated the ingested packets without significant symptomatology. Warning symptoms were present in 19 (11.8%) patients. Three patients (1.9%) presented with marked anxiety but none had cocaine in the urine sample. Fifteen (9.3%) body packers complained of colicky pain, and all underwent plain X-ray studies of the abdomen. Ten (6.2%) of them without radiologic signs of intestinal occlusion recovered with food deprivation and medical treatment, while five (3.1%) underwent laparotomy, three patients (1.9%) for gastric occlusion and two (1.2%) for ileal occlusion. Only one patient (0.6%) had warning symptoms and a urine screen positive for cocaine metabolites. In three cases of gastric occlusion, a gastrotomy was accomplished. In two cases of ileal occlusion, and in the patient with cocaine intoxication, packets were milked into the cecum, and some into the descending colon and rectum, until anal expulsion, by gentle pulling. Type 4 packets represent a new method of cocaine packaging, carrying the possibility of treating these body packers more conservatively than those transporting previous type of packets. Surgical approach to intestinal occlusion also may be conservative, because distal propulsion of the packets can be accomplished without entering the intestine.