ChaosHunter ®是一個獨立的軟體工具,旨在產生可讀的公式,您的數字數據模型的應用如下:
● 對金融市場生產買入/賣出信號
● 預測未來價值的時間序列,包括市場價格
● 建立科學數據模型
● 建立企業財務或銷售數據模型
● 預測實驗的結果等等
ChaosHunter要求您從電子表格或數據源輸入歷史文本文件或樣本數據。然後您需要選擇算術類別和您想使用的ChaosHunter的其他數學函數,之後它會產生的數值公式,讓您可以閱讀、理解、利用,甚至對外出售。ChaosHunter可用功能還包括神經網絡和混沌函數。
由ChaosHunter公式所產生的買入/賣出信號,可以被轉移到許多流行的交易平台,它使您能夠將您的模型與貿易數字的經紀公司。我們現在可以利用已經開發完成的界面,配合以下的平台就能讓您更輕鬆轉換ChaosHunter模型:
● NeuroShell® Trader Professional
● NeuroShell® DayTrader Professional
● Interactive Brokers Trader Workstation
● TradeStation®
● NinjaTrader®
● MetaTrader 4
● Wealth-Lab Pro®
● eSignal®
● Microsoft® Excel
Science and Trade Modeling
The ChaosHunter works by evolving formulas from basic building blocks - "atomic" functions like add, subtract, multiply, divide, sine, cosine, square root, etc. The user selects which of these functions will be in the pool of available functions, and the ChaosHunter evolves combinations that continually get better at solving the problem.
Solving the problem can mean predicting or classifying a time series for business and science users (primarily), or it can mean generating buy/sell signals for trading. Since the formulas that get generated are usually not esoteric like neural network formulas, you can show them to your boss, modify them, and insert them into other programs.
The ChaosHunter's atomic function set contains simple polynomials (e.g. a2 + b2), neurons, boolean functions (AND, OR, NOT), and many more. The neuron functions (if chosen) can combine to form unique neural net structures.
For our science and business users, you get models that have readable, understandable formulas to model the data. If the data is time series, it builds recurrent formulas necessary for most definitions of chaotic functions. Although it mostly predicts, you can classify as well if there are two classes, one of which can be described by positive numbers and the other by negative numbers.
The trading models it builds can be used standalone or in conjunction with NeuroShell Trader Professional or DayTrader Professional, and they are capable of investigating chaotic time series.
Traders can also fire the trading models the ChaosHunter makes from any number of trading products that you may have purchased before you found out how good NeuroShell Trader Professional is. The reason is that ChaosHunter makes formulas that can most likely be inserted into many of these systems. So you build the model (formula) based on text files exported from NeuroShell Trader or these other systems, then use the model in real time somewhere else. Of course, we have a much easier way to insert the models into NeuroShell...
Data Mining with Cubist
Cubist 資料採礦軟體工具
資料探勘(Data mining)是指從組織的資料庫中萃取資料的過程,這些資料通常被用來洞察該組織的營運模式和預設未發生之結果,以支援使用者做決策。
Cubist是 RuleQuest Research 公司開發的建立預測模型的工具。其內建的規則算法可幫助建立預測模型的輸出值,並與See5/C5.0產品互補。例如,See5/C5.0可能依據其百分比將數據分類為“高”、“中等”或“低”,而Cubist將會是輸出一個數字,如“7.3”。
Cubist是一個功能強大的工具,Cubist模型比那些一般的技術,如多元線性回歸得可以到更好的結果,同時也比神經網絡分析更容易理解。
DTREG 是一款完美的建模工具,適用於商業建模或建立多種醫學資料模型,例如性別、人種、婚姻狀況等! 從一組數據值中提取有用信息的過程被稱為「數據挖掘」(data mining)。這些數據可以用來創建模型並做出預測。目前的技術已經發展出許多預測模型,而如何選擇和應用最好的模型則是一種藝術。
DTREG實現強大的已開發預測建模方法。你可以使用決策樹模型、支援向量機(Support Vector Machine, SVM)、基因表示規劃法(Gene Expression Programming)、符號回歸(Symbolic Regression),K-means 分群法(k-means clustering)、判別分析法(discriminant analysis)、線性回歸(Linear Regression models)和(Logistic Regression models)。DTREG也可以進行時間序列分析和預測。
Classification and Regression Trees. DTREG can build Classification Trees where the target variable being predicted is categorical and Regression Trees where the target variable is continuous like income or sales volume.
DTREG 提供多種強大的預測建模
Multilayer Perceptron Neural Networks
Probabilistic Neural Networks
General Regression Neural Networks
RBF Neural Networks
GMDH Polynomial Neural Networks
Cascade Correlation Neural Networks
Support Vector Machine (SVM)
Gene Expression Programming - Symbolic Regression
Decision Trees
TreeBoost — Boosted Decision Trees
Decision Tree Forests
K-Means Clustering
Linear Discriminant Analysis (LDA)
Linear Regression
Logistic Regression
fuzzyTECH 為模糊邏輯與類神經模糊的模擬軟體,能結合類神經學習,以模糊邏輯得到解決方案,提供表列式、矩陣與FTL文字格式,讓使用者定義語言變數與模糊規則,規劃所需的模糊邏輯系統,適合複雜的模糊邏輯系統設計等應用。
fuzzyTECH軟體可以用於創建模糊系統,然後用神經網路和樣本數據對所創建的模糊系統進行訓練,最後可以針對不同的單片機或DSP硬體平臺自動產生相應的程序代碼,直接應用到系統的調試中去。
fuzzyTECH可以解決在單片機或DSP的有限資源中如何編制高效的應用程序以實現復雜的模糊控制算法的問題。它降低了實現模糊控制算法的難度,也加快了設計模糊控制算法的速度。fuzzyTECH設計電阻爐控制系統的例子,說明了它的有效性。
fuzzyTECH 提供三種版本,符合使用者不同工作平台的需求:
通用型(General Purpose fuzzyTECH Edition)
使用者在電腦中進行模糊邏輯的應用,包含專業版與線上版,特點如:
圖形式的編輯器與分析器使用介面
提供不同邏輯演算,完整互動式除錯分析
產生C/JAVA/M Code等模糊邏輯控制的原始副程式
與VB/VBA或C++/MFC整合,或DLL/OLE模組
與MS Office Excel/Access整合
與Matlab/Simulink、InTouch、FactoryLink、Genesis、WinCC 等軟體整合
線上版除包含專業版功能,並提供 TCP/IP、IPX/SPX、DDE、RS-232與SDI(User Define Interface)
等即時遠端監控分析介面。
單晶片控制(fuzzyTECH MCU Pack for Embedded Control)
應用於一般控制、數位信號處理與特定的模糊邏輯處理器,產生模糊邏輯控制的最佳化組合語言程式或C語言程式。支援8051/80251、Motorola 68HC05/68HC08、Microchip、Mitsubishi、Intel MCS-96、Siemens、TMS-320與SAE81C99A等系列控制器。
fuzzyTECH MCU-HC05/08 Edition: Supports all microcontrollers of the 68HC05 and 68HC08
families from Motorola.
fuzzyTECH MCU-HC11/12 Edition: Supports all microcontrollers of the 68HC11xx and 68HC12xx
families from Motorola. Utilizes the special fuzzy logic instruction set of the HC12. RTRCD functionality included.
fuzzyTECH MCU-MP Edition: Supports all microcontroller families of Microchip Technologies Inc.
(PIC16C5X, PIC16CXX, PIC17CXX).
fuzzyTECH MCU-51 Edition: Supports all microcontrollers of the 8051 and 80251 families. Special libraries for
80517 included. RTRCD functionality included.
fuzzyTECH MCU-96 Edition: Supports all microcontrollers of the MCS-96 family from Intel (8096, 80C196, ...).
RTRCD functionality included.
fuzzyTECH MCU-166 Edition: Supports all microcontrollers of the C16x family from Siemens.
RTRCD functionality included.
fuzzyTECH MCU-320 Edition: Supports all digital signal processors (DSP) of the TMS-320 C2x/3x/4x and
5x families from Texas Instruments.
可程式控制(fuzzyTECH IA Editions for Industrial Automation)
提供模糊邏輯函數方塊,整合至PLC可程式控制器與程序控制系統,以歸屬函數、MIN-MAX推論、反模糊化方法與即時監控功能,完成模糊邏輯系統。支援Klockner-Moeller、Foxboro/Eckardt IAS等程序控制。
fuzzyTECH IA-S7 Edition: For Siemens SIMATIC S7™ (300 and 400) PLCs.
fuzzyTECH Professional Edition: For Structured Text by the I...