A Cognitive IoT Learning Models for Agro Climatic Estimation Aiding Farmers in Decision making

Figure 1:

Figure 2:

Figure 3:

Figure 4:

Performance Metrics Evaluation for the Proposed Architecture using Testing Datasets

Sl.no	No of batches	No of Epochs	Precision (%)	Recall(%)	F1-Score
01	160	40	96.4%	96.1%	96.3%
02	160	80	96.43%	96.31%	96.3%
03	160	120	97.6%	97.41%	97.52%
04	160	160	97.53%	97.12%	97.32%
05	160	200	96.4%	96.3%	96.2%
06	160	240	96.3%	96.2%	96.1%
07	160	280	96.2%	96.1%	96.1%

Evaluation of Various Models for Predicting Crop Yield Productivity with Dropout Rate of 0_2

Algorithm	Performance Metrics(%)		recall	Specificity	F1-score
	Accuracy	Precision
LSTM	82.5	83.5	83.4	84.1	84.2
LSTM+PSO	86.3	86.2	85.8	86.1	85.4
LSTM+GA	86.6	86.4	86.2	87.3	85.9
LSTM+WOA	88.4	87.3	86.8	87.1	86.9
LSTM+SSO	88.5	89.1	88.8	88.4	88.5
LSTM+SHO	91.6	89.7	89.4	89.4	88.4

Benchmarking assessment of the various methodologies in identifying the crop-yield productivity with the drop-out=0_6

Algorithm	Performance Metrics(%)		recall	Specificity	F1-score
	Accuracy	Precision
LSTM	83.5	83.6	83.5	84.0	84
LSTM+PSO	88.3	86.3	85.9	86.0	85.4
LSTM+GA	88.3	86.5	86.3	87.4	85.9
LSTM+WOA	89.6	87.4	86.9	87.2	86.9
LSTM+SSO	89.9	89.0	88.9	88.5	88.5
LSTM+SHO	90.8	89.8	89.5	89.5	88.4
PROPOSED MODEL	97.6	96.9	96.7	96.6	96.5

Performance Metrics Evaluation for the Proposed Architecture using Verification/Validation Datasets

Sl.no	No of batches	No of Epochs	Precision (%)	Recall(%)	F1-Score
01	170	40	96.5%	96.%	96.3%
02	170	80	96.45%	96.3%	96.3%
03	170	120	97.5%	96.9%	97.51%
04	170	160	97.3%	97.1%	97.31%
05	170	200	96.45%	96.30%	96.4%
06	170	240	96.30%	96.20%	96.20%
07	170	280	96.20%	96.10%	96.17%

Validation/Verification Accuracy Performance using the no of batches =160

Sl.no	No of batches	No of Epochs	Validation /Verification Accuracy (%)
01	160	40	96.34%
02	160	80	97.6%
03	160	120	98.55%
04	160	160	98.35%
05	160	200	98.21%
06	160	240	98.3%
07	160	280	98.2%

Benchmarking assessment of the various methodologies in identifying the crop-yield productivity with the drop-out=0_4

Algorithm	Performance Metrics(%)		recall	Specificity	F1-score
	Accuracy	Precision
LSTM	84.3	83.6	83.5	83.5	84
LSTM+PSO	87.6	86.3	85.9	88.3	85.4
LSTM+GA	87.5	86.5	86.3	88.3	85.9
LSTM+WOA	89.5	87.4	86.9	89.6	86.9
LSTM+SSO	89.4	89.0	88.9	89.9	88.5
LSTM+SHA	91.0	89.8	89.5	90.8	88.4
PROPOSED MODEL	97.3	96.9	96.7	96.9	96.5

Performance Metrices

SL.NO	Performance Measures	Expression
1	Accuracy	TP+TNTP+TN+FP+FN {{TP + TN} \over {TP + TN + FP + FN}}
2	Recall	TPTP+FN×100 {{{\rm{TP}}} \over {{\rm{TP}} + {\rm{FN}}}}\, \times 100
3	Specificity	TNTN+FP {{TN} \over {TN + FP}}
4	Precision	TNTP+FP {{TN} \over {TP + FP}}
5	F1-Score	2.Precison ∗ RecallPrecison + Recall 2.{{Precison\, * \,{Recall}} \over {Precision\, + \, {Recall}}}

Training Accuracy Performance using the no of batches =160

Sl.no	No of batches	No of Epochs	Testing Accuracy (%)
01	160	40	96.36%
02	160	80	97.65%
03	160	120	98.55%
04	160	160	98.35%
05	160	200	98.21%
06	160	240	98.3%
07	160	280	98.2%