Fair evaluation of classifier predictive performance based on binary confusion matrix

Vanacore, Amalia; Pellegrino, Maria Sole; Ciardiello, Armando

doi:10.1007/s00180-022-01301-9

Cited by 11 publications

(3 citation statements)

References 34 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…A confusion matrix is used to evaluate the model's performance in a classification problem. Finding three crucial factors, namely accuracy, sensitivity, and specificity, is done by using the confusion matrix's elements [25].…”

Section: Performance Measuresmentioning

confidence: 99%

Crime Classification Using GRU and 1CNN Techniques

S. Jeya Selvakumari, V. Joseph Peter

2023

tjjpt

View full text Add to dashboard Cite

Criminal activity is a widespread social problem that affects a nation's standard of living, rate of economic development, and international standing. Numerous crimes have proliferated worldwide in recent years. It has a variety of effects on people of all types. Classification of crime is necessary to create a safe society. Many researchers classify the crimes using machine learning techniques with a common dataset, which presents numerous computational opportunities and challenges. Using the learning capabilities of several deep learning architectures, we are trying to classify the criminal activity from the real-time dataset of particular city. In the proposed work, Gated Recurrent Unit (GRU) and Convolutional Neural Network (1CNN) deep learning architecture are used for crime classification. We have conducted performance comparison for classifying the crime with real-time dataset. The effectiveness of the aforementioned deep learning models was assessed by Training and Validation Loss curve. It was discovered that 1CNN models achieve better result compared to GRU. 1CNN produced 97% accuracy.

show abstract

Section: Performance Measuresmentioning

confidence: 99%

Crime Classification Using GRU and 1CNN Techniques

S. Jeya Selvakumari, V. Joseph Peter

2023

tjjpt

View full text Add to dashboard Cite

show abstract

“…''Positive'' indicates when the ramp event is forecasted to occur, and ''Negative'' refers to when the ramp event is forecasted to not occur. By combining T or F and P or N according to whether the forecasting and the actual measurement match, the forecasting result is represented as a total of four cases: TP, FN, FP, and TN [28].…”

Section: Performance Validation Metric For Ramp Event Detection 1) Co...mentioning

confidence: 99%

An Enhanced Performance Evaluation Metrics for Wind Power Ramp Event Forecasting

Yu,

Hur

2023

IEEE Access

View full text Add to dashboard Cite

As wind power is volatile and intermittent, an increase in the proportion of wind power generation may affect the security of power systems. Accurate wind power generation and ramp forecasting are crucial for reducing operating costs and maintaining the power balance. This study proposed an enhanced performance evaluation metric for wind power ramp event forecasting, and analyzed the forecasting results using this metric. Experimental ramp event forecasting was conducted on wind farms located in Jeju, South Korea by employing an indirect forecasting method for wind power output forecasting. Considering the evaluation of forecasting performance, the accuracy (ACC) was observed to be 0.86 for Case #1 and 0.90 for Case #2. The results highlight the advantages of the proposed metric over the widely used confusion matrix for performance evaluation, which has more detailed and visually based analytical capabilities. The simulation demonstrates that the proposed approach is poised to make a noteworthy contribution in the advancement of tool development, enabling real-time curtailment forecasting and facilitating well-informed decision-making in high wind power scenarios.INDEX TERMS Performance evaluation metric, power system security, ramp event forecasting, wind power.

show abstract

“…Hasil prediksi dari proses training dan testing ditunjukkan pada gambar berikut: Accuracy, sensitivity (recall), precision, dan F1-score adalah beberapa ukuran kinerja prediktif yang umum digunakan berdasarkan confusion matriks biner [21]. Setiap ukuran berkaitan dengan aspek kinerja tertentu, sehingga ukuran kinerja yang sesuai untuk masalah yang dihadapi umumnya dipilih sesuai dengan aspek kinerja yang akan diselidiki.…”

Section: Evaluasi Hasilunclassified

Prediksi Guru Kemungkinan Tetap Bekerja di Sekolah Al Uswah Surabaya Menggunakan Machine Learning

Edris Effendi,

Yuadi,

Puspitasari

2023

jidt

View full text Add to dashboard Cite

Penelitian ini bertujuan untuk memprediksi potensi tetap bekerjanya guru di Sekolah Al Uswah Surabaya, sehingga dapat meminimalisir tingkat turnover. Peneliti telah menggunakan algoritma klasifikasi untuk mengetahui algoritma yang paling cocok dalam memprediksi turnover guru dan diproses dengan aplikasi orange data mining. Berdasarkan tabel hasil prediksi maupun confusion matrix, menghasilkan rekomendasi bahwa algoritma yang paling bagus performanya adalah Logistic Regression. Tingkat Presicion untuk perbandingan data training dan testing 80:20 mencapai 80,8%, lebih tinggi dibanding tiga algoritma lainnya yang di bawah 80%. Melalui penelitian ini dapat memperjelas bahwa untuk studi kasus prediksi turnover karyawan, dapat menggunakan parameter Presicion. Melalui hasil penelitian dapat membantu lembaga pendidikan dalam merekrut guru yang memiliki peluang bisa bertahan lebih lama dengan memanfaatkan beberapa atribut standar dari biodatanya.

show abstract

Fair evaluation of classifier predictive performance based on binary confusion matrix

Cited by 11 publications

References 34 publications

Crime Classification Using GRU and 1CNN Techniques

Crime Classification Using GRU and 1CNN Techniques

An Enhanced Performance Evaluation Metrics for Wind Power Ramp Event Forecasting

Prediksi Guru Kemungkinan Tetap Bekerja di Sekolah Al Uswah Surabaya Menggunakan Machine Learning

Contact Info

Product

Resources

About